Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Appendix F Connected and Automated Vehicle Technology Impacts on Future Interstate Highway System Steven E. Shladover EXECUTIVE SUMMARY This appendix provides an overview of the implications that development and deployment of connected and automated vehicle (CAV) technology could have for the future of the Interstate Highway System. It begins with an introduction describing the range of connected vehicle (CV) technologies and applications and of automated vehicle (AV) alternatives. The AV alter- natives are characterized on the basis of their level of automation, whether their operations are autonomous or cooperative, and on the operational design domains (ODDs) within which they are capable of operating at any stated level of automation. The levels of automation are defined based on the SAE classifications that specify the relative roles of the driver and the driving automation system. The ODD represents the range of roadway classifications, speeds, traffic, and weather conditions in which any specific AV system is able to operate, and when that is combined with the levels of automation, it represents the great diversity of AV system concepts of operation and capabilities. Many aspects of the technological development and public implemen- tation of the CAV technologies are shown to be highly uncertain. Because CAV systems integrate elements from the information technology, vehicle technology, and infrastructure industries, the implementation of these sys- tems is conditioned by the contrasts in the timescales within which these industries evolve. The real-world implementation times are much more likely to be governed by the slowest of these than by the fastest. The deployment patterns are also likely to vary substantially among regions of the country, 351
352 NATIONAL COMMITMENT TO THE INTERSTATE HIGHWAY SYSTEM and so a single national deployment profile cannot be assumed. Given these uncertainties, decisions about future roadway infrastructure should not be finely tuned to specific assumed technology deployment schedules, but rather need to be robust with respect to a wide range of possible CAV deployment patterns. The CAV technologies will influence both the supply and demand sides of the transportation system, with somewhat different implications for passenger and freight transport. The expected influences of CAV (and related information technologies) on passenger and freight travel demand are described separately. There will be both positive and negative influences on the number of trips to be taken, as well as on the character of the trips, and because of this diversity of impacts, it is very difficult to make reliable predictions about the net impacts. The influences of CAV technologies on the supply side of road transportation operations are also discussed, in this case with somewhat less uncertainty based on results from early technologi- cal experiments that show the potential to increase highway capacity and improve traffic flow dynamics. The appendix concludes with estimates of realistic expectations for the changes that CAV technologies could make to interstate highway opera- tions within the next 10, 20, and 50 years. At the 10-year horizon, the early impacts of CV technology should be felt, but that will be too early to see significant impacts from AV technology. By the 20-year point, the lower-level AV technologies should be in use on enough vehicles that their impacts should become evident, but these will still not produce revolution- ary changes to the design or operation of the Interstate Highway System. The 50-year mark is more challenging to predict because of the many un- certainties involving the rate of maturation of the AV technologies and the extent to which they will be embraced by the public for widespread use. This means that 50 years from now the transportation system could on the one hand be only modestly different from today or it could on the other hand be largely transformed into something more highly automated and connected than todayâs system. INTRODUCTION OF CONNECTED AND AUTOMATED VEHICLE TECHNOLOGIES Modern information technology has been gradually entering the road transportation system over the past two decades under the label of intel- ligent transportation systems (ITSs). These represent combinations of sen- sor systems (on both vehicles and infrastructure), computer hardware and software to process the sensor data and make decisions, communication systems to exchange information among vehicles and between vehicles and the roadway infrastructure, actuators to command vehicle motion changes;
APPENDIX F 353 and humanâmachine interfaces to exchange information with drivers and travelers. In recent years, attention has been focused specifically on CV systems and AV systems, and on their combination (CAV systems). The CV systems exchange information among vehicles and between vehicles and the road- way infrastructure, while the AV systems relieve drivers of some or all of the tasks associated with controlling the motions of vehicles. The specific CV and AV technologies are introduced in this section to provide a foundation for the discussion that follows about their implications for the future of the Interstate Highway System. Connected Vehicle Systems For at least the past 15 years, the primary focus of the U.S. Department of Transportation (U.S. DOT) ITS program has been on CV systems, although this has been done under several different labels (Intelligent Vehicle Initia- tive, IntelliDrive, and now Connected Vehicles). The basic concept is largely the same, that wireless communications among vehicles (vehicle-to-vehicle [V2V]) and between vehicles and the roadway infrastructure (vehicle- to- infrastructure [V2I] and infrastructure-to-vehicle [I2V]) and between vehicles and other entities such as pedestrians and bicyclists (vehicle-to- everything [V2X]) can enable the transportation system to function more effectively as an integrated system. The primary emphasis of most of these efforts has been on improv- ing safety by use of one specific wireless technology, 5.9-GHz dedicated short-range communication (DSRC). However, that scope has broadened considerably to encompass other wireless technologies and improvements in other transportation system measures of effectiveness. V2V connectivity can enable applications such as â¢ Cooperative collision warnings and hazard alerts, as tested in the Safety Pilot Model Deployment; â¢ Cooperative collision mitigation or avoidance, incorporating active braking; â¢ Cooperative adaptive cruise control, with tighter vehicle following control than conventional adaptive cruise control and enhanced traffic flow stability; â¢ Close-formation automated platooning, enabling aerodynamic drafting and lane capacity increases; â¢ Automated maneuver negotiation at merging locations or intersec- tions; and â¢ Transit bus connection protection.
354 NATIONAL COMMITMENT TO THE INTERSTATE HIGHWAY SYSTEM All but the last of these are time-critical and safety-critical applications that need very low-latency and high-reliability communications. For most of these applications, the communicated data are used to augment the data acquired by onboard remote sensors, which remain the primary source of data about time-critical and safety-critical conditions. I2V connectivity can provide â¢ Traffic signal status information in real time for in-vehicle display, signal violation warning, or green wave speed advisories to drivers; â¢ Traffic and weather condition information and real-time routing advisories to drivers; â¢ Fleet management functions of vehicle routing and scheduling; â¢ Access control to closed facilities; â¢ Variable speed limits and advisories directly to drivers or their vehicles (I2V cooperative adaptive cruise control); â¢ End-of-queue warnings; and â¢ Active support for lane guidance. V2I connectivity can enable â¢ Vehicle probe data applications providing detailed traffic informa- tion (speed, volume, travel time, queue length, stops) or road surface condition information (pavement roughness or slippery conditions); â¢ Mayday and concierge services (such as OnStar); â¢ Electronic toll collection and parking payments; â¢ Traffic signal priority requests; and â¢ Vehicle status information for fleet management (especially for transit and trucking fleets). CVs can rely on a variety of wireless communication technologies for their connectivity. In the earlier years of the Vehicle-Infrastructure Integra- tion (VII) program within the U.S. DOT, attention was focused entirely on 5.9-GHz DSRC technology, since that special-purpose wireless technology was designed specifically for the mobile environment and it was thought that it could serve all mobile communication needs. Since then, it has be- come evident that a wider range of wireless technologies, with differing strengths and weaknesses, could be used to support ITS applications: â¢ The 5.9-GHz DSRC is a special WiFi-like technology designed for road transportation applications, but using a licensed and pro- tected spectrum. It has a unique ability to support time-critical and safety-critical messages over a limited range. Since it is spe- cific to the road transportation environment, it benefits less from
APPENDIX F 355 commercial developments and economies of scale than some of the other technologies. The National Highway Traffic Safety Adminis- tration (NHTSA) has released a Notice of Proposed Rulemaking to create a new Federal Motor Vehicle Safety Standard (FMVSS) 150 that would require all new vehicles to be equipped with this radio technology starting in the early 2020s so that the information that they broadcast periodically (10 times per second) can be used by other vehicles to predict potential crashes and alert drivers to avoid those crashes (NHTSA 2017). â¢ WiFi can be used, where available, to support some ITS functions, but it has relatively long connection latency and is susceptible to delays and packet losses when the channel is congested, and so its dependability is inadequate for critical information. â¢ Cellular communications include 4G LTE and WiMAX technolo- gies and their future generations, generally termed 5G cellular. The infrastructure side of the current generation of this system is essentially ubiquitous in built-up areas, and so it does not need to be provided by public agencies, but users on both the vehicle and infrastructure sides need to pay the network operators for data usage. The development of 5G cellular is under way, but its real capabilities are not yet clearly defined and its deployment timescale remains uncertain. â¢ Satellite communication systems can be used in remote areas that lack cellular service, but they face significant cost, bandwidth, and latency limitations, which mean that they are not suitable for all applications. â¢ Bluetooth can provide only very short-range and low-bandwidth service to support some applications. State and local transportation agency issues come to the forefront for DSRC rather than the other alternatives because the implementation of the infrastructure elements of this system will most likely depend on actions by these agencies (at least providing access to their field devices and probably funding as well). DSRC is needed for the V2V safety and I2V intersection safety applications, and so these are the ones that will require involvement by the transportation agencies. The transportation agencies will also need to be involved in the I2V and V2I applications that support or influence transportation system manage- ment, regardless of the wireless technology that is selected. As the owners and operators of the transportation management centers, they will be the users and, in many cases, the providers of the data, so they need to be actively engaged in the development, deployment, maintenance, and opera- tion of these information systems.
356 NATIONAL COMMITMENT TO THE INTERSTATE HIGHWAY SYSTEM Automated Vehicle Systems AV systems have had a considerably longer and more irregular history than CV systems, with many ups and downs since the concept of road vehicle automation was introduced by Norman Bel Geddes in the General Motors Futurama exhibit at the 1939â1940 New York Worldâs Fair (Bel Geddes 1940). The first wave of research and development on automated road ve- hicles was undertaken by General Motors and RCA beginning in the 1950s (Bender 1991), and the second wave was led by Robert Fenton at The Ohio State University from 1964 to 1980 (Fenton and Mayhan 1991). The third wave was initiated with the founding of the PATH Program by Caltrans and the University of California in 1986 (Shladover 1990), reached a cli- max with the research and demonstration work of the National Automated Highway Systems Consortium from 1994 to 1998 (Rillings 1997), and con- tinued until 2003, when Caltrans and PATH did joint demonstrations of au- tomated bus platoons and truck platoons. Parallel activities were initiated in Europe and Japan in the mid-1980s, including pioneering research on the application of video image processing for driving scene recognition by Ernst Dickmanns (2002) and the wide-ranging research and demonstration work conducted under the PROMETHEUS program in Europe (Glathe 1994) and the Super-Smart Vehicle Systems (SSVS) program in Japan (Tsugawa et al. 1992). The fourth wave, which continues today, began with the DARPA Challenges in the 2004â2007 period and the ensuing work by Google, with its first public announcement in 2010. There have been many journalistic reviews of the hype surrounding this wave but relatively few sober assess- ments (Anderson et al. 2016; Shladover 2016). One of the primary current challenges in the AV domain is managing the unrealistic expectations of the general public, elected officials, and some transportation professionals. Media coverage and Internet chatter has mis- led many people about what capabilities can be achieved within the coming years and decades, and the imprecise and inaccurate vocabulary used to discuss AVs adds to the confusion. The words âdriverless,â âself-driving,â and âautonomousâ are frequently applied to systems that still depend very heavily on human engagement in the driving task, blurring important distinctions among the capabilities of different systems. The companies that are active in this space are competing strongly for media attention and public image, and so they have been making misleading statements and encouraging media people to extrapolate beyond the reality of what is actually likely to be deployed for public use. It is important to begin consideration of AVs by recognizing the great diversity of AV applications and concepts of operation and the large differ- ences in when they will become available for use by the public. The highest levels of automation have the potential to revolutionize the transportation
APPENDIX F 357 system in a variety of ways, but there are large uncertainties about how far in the future those will become available. On the other hand, the lower lev- els of automation are imminent and transportation officials need to under- stand their limitations and their implications for the transportation system. There are three important dimensions of classification of AV systems: â¢ Level of automation capability, â¢ Distinction between autonomous (unconnected) and cooperative (connected) implementations, and â¢ ODD. Although most of the earliest automation systems to become available, which provide low levels of automation of driving functions, are autono- mous, over the longer term as higher levels of automation are developed, it will be increasingly important for the automation systems to be cooperative (cooperation among vehicles and between vehicles and the roadway infra- structure) in order to produce transportation system benefits. The ODD is the combination of specific conditions under which a driving automation system is designed to function, including driving modes, roadway types, traffic conditions, speed range, geographic locations (boundaries of digital maps), weather and lighting conditions, and availability of necessary sup- porting infrastructure features including condition of pavement markings and signage, and so forth. SAE International (formerly the Society of Automotive Engineers) has developed a five-level classification of automated driving systems, which is very useful for distinguishing the capabilities that will be available at each level (SAE International 2016). Without belaboring the fine points of the definitions that distinguish the five levels from each other, Table F-1 lists examples of the types of systems that would fit within each level and the roles that the driver would have with each of these systems. The transportation system impacts of these different levels of driving automation system will differ greatly, and so it is important to consider them separately. It is also vital for transportation agency decision makers to make their decisions based on realistic predictions of the timing of the availability of these varying levels of automation. Level 1 driver assistance systems are already on the market on a variety of vehicles, although they still represent a small fraction of the number of vehicles sold. Adaptive cruise control (ACC), the most important Level 1 system, was first introduced on a production car in Japan in 1995 and first available in the United States in 2000, but it is still not in widespread use, even though this is a feature that is regarded very favorably by people who have used it. Wayland (2015) reported that ACC was installed on 2.2 per- cent of new vehicles worldwide in 2014, and that number was projected to
358 NATIONAL COMMITMENT TO THE INTERSTATE HIGHWAY SYSTEM increase to only 7.2 percent by 2020. Level 2 partial automation systems have recently been introduced on high-end vehicles, and will be introduced on premium vehicles from more manufacturers within the next few years. Both Level 1 and Level 2 systems provide driving comfort and convenience, but they require that the driver continuously monitor the driving environ- ment for hazards and be prepared to resume control immediately when the system encounters situations it cannot handle. These are not expected to have significant impacts on the transportation system while they exist in limited numbers. As the market penetration grows, the Level 1 systems should produce some safety increase but the safety implications of Level 2 systems are uncertain because of the likelihood that drivers will misuse them by diverting their attention to other activities while the systems are in use. There is already significant evidence of this misuse in experimental find- ings from NHTSAâs initial evaluations of the systems (Blanco et al. 2015), Googleâs experience when loaning its test vehicles to company employees not directly involved in the development of the vehicles, and YouTube videos1 posted by members of the general public. This type of misuse will create new safety problems and has already cost the lives of several Tesla âautopilotâ drivers. The primary safety benefits for the foreseeable future from use of CAV technology are likely to come from Level 1 and Level 0 automation systems. 1 See https://youtu.be/2AM9Qqcir6k and https://youtu.be/zY_zqEmKV1k. TABLE F-1 Summary Descriptions of SAE Levels of Automation Level Example Systems Driver Roles 1 Adaptive cruise control OR lane-keeping assistance Must drive other function and monitor driving environment 2 Adaptive cruise control AND lane- keeping assistance Traffic Jam Assist for freeway (Mercedes, Tesla, Infiniti, Volvo, and so forth) Parking with external supervision Must monitor driving environment (system nags driver or deactivates itself to try to ensure this) 3 Traffic jam pilot May read a book, text or Web surf, but be prepared to intervene when needed 4 Highway driving pilot Closed campus âdriverlessâ shuttle âDriverlessâ valet parking in garage May sleep, and system can revert to minimum risk condition if needed 5 Ubiquitous automated taxi (even for children) Ubiquitous car-share repositioning system Can operate anywhere, with no driver needed
APPENDIX F 359 At Level 0, the systems do not perform dynamic driving tasks on a sustained basis, but they can warn drivers about hazards or even intervene for a brief time to avoid or mitigate an imminent crash (automated emergency braking or lane departure prevention). At Levels 0 and 1, the driver must remain fully engaged in the dynamic driving task, so that the vigilance of the sys- tem and the driver support each other to produce a higher overall level of vigilance with regard to hazards. At the higher levels of automation, driver vigilance is inevitably reduced, and so the vigilance requirements on the system become significantly more severe. Level 3 conditional automation systems will provide higher levels of driver comfort and convenience by allowing the driver to temporarily turn attention away from driving to engage in other activities, but the driver still needs to be available to retake control within a few secondsâ notice when the system reaches the limits of its capabilities. It is not yet clear whether it will be possible to implement a driverâvehicle interface that can success- fully manage these transitions and prevent the driver from âtuning outâ so seriously that she or he is unable to intervene when needed. Although many vehicle manufacturers have steered clear of Level 3, Audiâs new Traffic Jam Assist feature provides the Level 3 functionality of bringing the vehicle to a stop if the driver fails to intervene (Audi 2017; Markoff 2017). Level 4 high automation includes a diverse collection of capabilities that need to be considered individually. These systems can replace drivers completely (not requiring driver interventions), but only under specific limited conditions, and those limitations can vary widely from system to system: â¢ Automated valet parking systems will park cars in parking lots or garages after the driver has exited the vehicle, making it possible to squeeze them into smaller parking spaces in areas where land is expensive. The first systems require continuous supervision by the driver using a key fob or software application on a smart phone or tablet. In the next few years, systems will be able to operate out- side the driverâs line of sight, in suitably equipped parking facilities that limit hazards and provide supplementary communication and sensing capabilities. Eventually, automated valet parking will be extended to any parking facility. â¢ Automated buses on special transitways will be developed as cost- effective alternatives to light-rail transit on high-volume urban routes (Shladover 2000). The automation technology will provide a rail-like quality of service and the ability to fit within a narrow right-of-way through accurate steering control, but at much lower cost than a rail system. The physically constrained environment of the transitway and the ability of the transit operator to equip it
360 NATIONAL COMMITMENT TO THE INTERSTATE HIGHWAY SYSTEM with cooperative infrastructure elements should make it possible to start implementing this capability within this decade. â¢ Automated trucks on dedicated truck lanes are another high-value niche application of automation that should be possible within the decade by restricting access to those lanes to trucks (Shladover 2001). By excluding light-duty vehicles and vulnerable road users from coexisting with the trucks and operating within a physically constrained highway environment, the most challenging hazards can be eliminated, to enable Level 4 highway automation within the decade if the truck-lane infrastructure can be made available. There is a strong economic incentive to truck owners and operators to implement this technology because of its significant fuel-saving potential. â¢ Automated low-speed shuttles in campuses or pedestrian zones have been the focus of much attention in Europe through the CityMobil2 project,2 and several small companies have been developing vehicles for this type of application. Google (now Waymo) also shifted its attention with the announcement about its âpod carâ emphasis since 2014. The European work has depended on certification of the infrastructure where the vehicle travels, with special design features to limit the interactions with other road users and to ensure clear fields of regard for the vehicle sensors that need to detect hazards. Based on the reduced speeds and infrastructure restrictions, the first of these vehicles could be operational without drivers onboard before the end of the current decade (although they would still be supervised remotely from an operations center). â¢ Automated passenger cars on limited-access highways are likely to be the most broadly applicable Level 4 automation systems. These automation systems will probably initially (in the 2020â2025 time frame) only be usable under certain traffic conditions (such as low-speed traffic jams or high-speed operations in light traffic) or in lanes that are restricted to vehicles that are equipped for auto- mation and/or V2V communication capabilities, analogous to the automated highway system concepts that were developed by the NAHSC in the 1990s (Rillings 1997). These restrictions will facili- tate safety before the hazard detection technology has advanced sufficiently to handle all highway driving hazards, which is likely to be closer to 2030. Level 5 full automation would enable a vehicle to drive itself anywhere and under any conditions in which a normal human driver would be able to 2 See http://www.citymobil2.eu/en.
APPENDIX F 361 drive. This is the concept that captures the public imagination by allowing full âelectronic chauffeurâ service, including â¢ Electronic taxi service for people who are not able to drive (too old, too young, physically impaired), â¢ Shared vehicle fleet repositioning so that shared vehicle concepts can be economically efficient, and â¢ Driverless urban goods pickup and delivery. These applications are the ones that could have revolutionary impacts on travel behavior and urban form by eliminating the disutility of travel time, decoupling parking locations from travelersâ origins and destinations, fa- cilitating vehicle sharing as well as ride sharing, and breaking down the boundaries between public and private transportation. However, the tech- nological problems that need to be solved before this can become reality are extremely daunting, and so these capabilities are not likely to become available for many decades. For each of these levels of automation, it is important to be conscious of the time between market introduction and widespread use. Even if a new feature is implemented on every newly manufactured vehicle, it will take close to 20 years for the vehicle fleet to turn over sufficiently that it will be found on the large majority of the vehicles on the road. Most technol- ogy dissemination on private vehicles is a lot slower than that because the technology is introduced only on the high-end vehicles in limited quantities before manufacturers scale up to larger-volume production, and then it depends on private individualsâ voluntary vehicle purchase decisions; so it is important to allow additional decades for the technologies to propagate through the vehicle fleet. If the government mandates inclusion of specific features on all new ve- hicles it can accelerate their market introduction. For example, when NHTSA mandated the inclusion of seat belts on all new vehicles, there was a gradual phase-in period so that it took 6 years for more than 90 percent of new ve- hicles to have seat belts, but it took 22 years for 90 percent of the vehicles on the road to be equipped with seat belts. In the absence of a government mandate, the rate of propagation depends on market forces. Past experience provides examples of the time it took from first market introduction to in- stallation on 90 percent of new vehicles for major automotive technologies that we now take for granted as standard equipment (Jutila and Jutila 1986): â¢ Air conditioning, 20 years; â¢ Automatic transmission, 27 years; â¢ Power steering, 21 years; and â¢ Disc brakes, 10 years.
362 NATIONAL COMMITMENT TO THE INTERSTATE HIGHWAY SYSTEM Demand-Side Impacts The CAV technologies are likely to have diverse impacts on the demand for travel, and the net impact is likely to be challenging to estimate because of the large uncertainties on both the positive and negative expected impacts. These demand effects are expected to include â¢ Reductions in the need to travel, with potential for substituting telecommunications activities for travel; â¢ Changes in trip scheduling, with better information promoting bet- ter choices to avoid the worst congestion and safety challenges; â¢ More efficient selection of routes and modes based on better infor- mation about all viable alternatives; â¢ Reduction in disutility of travel time, encouraging realization of latent demand and potentially inducing new travel demand through locational changes; â¢ Increased efficiency and improved quality of service by trucking, encouraging freight modal shift toward trucking; â¢ Improved transit service quality, encouraging passenger mode shifts away from private personal vehicles and toward transit; and â¢ Electronic chauffeuring providing affordable mobility for travelers who cannot drive, encouraging them to travel more than before. Some of these impacts are likely to reduce vehicle-miles traveled (VMT) and some are likely to increase it, which makes it particularly challenging to estimate the net VMT effect. If the mobility enhancement effects dominate, VMT is likely to increase unless ridesharing in automated jitney services becomes the preferred mode of urban and suburban transport (in which case VMT could decrease). Supply-Side Impacts The CAV technologies are likely to have even larger supply-side effects, producing significant changes in multiple aspects of traffic operations. These changes are likely to affect virtually all of the significant measures of effectiveness by which traffic operations are measured, including safety, travel times, congestion, energy use, emissions, and travel comfort and convenience. In most cases, these effects will depend heavily on whether the CV and AV technologies are implemented individually or in combination. These effects are expected to include â¢ Changes in traffic flow stability based on differences in vehicle fol- lowing dynamics,
APPENDIX F 363 â¢ Changes in highway lane capacity based on differences in vehicle following gaps, â¢ Increases in highway bottleneck throughput based on more re- sponsive traffic management and ability to implement situation- dependent speed control, â¢ Reduction in traffic disturbances from lane drops and entrance and exit ramp flows through coordinated vehicle merging, â¢ Improved ability to manage incidents based on higher-fidelity in- formation for incident responders and for travelers, and â¢ Improved multimodal corridor management in urban corridors through enhanced information and control mechanisms. Mode-Specific Impacts The later sections of this appendix address the freight and passenger move- ment impacts separately because there are some significant differences be- tween them on both the supply and demand sides. These distinctions can be lost if they are aggregated at too gross a level. CENTRAL IMPORTANCE OF UNCERTAINTY IN CONSIDERING FUTURE TECHNOLOGY IMPACTS Forecasting the development of information technology is fraught with uncertainty, since this is a domain in which the âdisruptersâ of today become the âestablishmentâ of tomorrow and the cycles of technological upheaval occur frequently. Silicon Valley thrives on revolutionary change and destructive innovation, with the expectation that the life cycle of each technological revolution is likely to be only a few years, up to maybe a couple of decades for the most durable ones. The contrast with planning for transportation infrastructure, with its expected functional lifetimes of many decades, is stark. When we enter the realm of CAVs, we are dealing with a complicated mix of information technology, vehicle technology, and infrastructure tech- nology. These three domains are radically different in their product life cycles and investment horizons. Although information technology is on the fast track, with product functional lifetimes measured in months, vehicles are designed for functional lifetimes of years and civil infrastructure must be designed to last for decades. These differences are to a large extent inherent in the capital intensity of the products and production processes associated with each industry, so we should not expect vehicles and civil infrastructure to change as rapidly as mobile phones and their application software.
364 NATIONAL COMMITMENT TO THE INTERSTATE HIGHWAY SYSTEM The most important principle to take away from this observation is that decisions about the future of our transportation infrastructure need to be robust with respect to technological uncertainty, rather than being highly tuned to specific predictions about technological outcomes. Each technol- ogy forecast should be viewed as a fuzzy estimate, with a wide range of possible outcomes that could range from much faster and higher capabili- ties to much slower and lower capabilities than the nominal prediction. Regardless of where reality eventually falls within this range of possible outcomes, transportation decision makers need to make sure that they have not locked themselves into an untenable situation by becoming too heavily dependent on one specific outcome, requiring a specific technological capa- bility to become available for use by a specific future date in order to justify the viability of an infrastructure investment or to meet a mobility need. The more optimistic predictions of change in information technology also need to be tempered by the realization that the vehicle and infra- structure technologies cannot change as rapidly, and so these are likely to become the pacing items for the rate of change in the road transportation system of the future. Although CV and AV have been grouped together for consideration here, their situations are quite different in terms of technological maturity and uncertainty. The CV technology has been under development for al- most two decades, with close coordination among the government and the vehicle and communications industries. Assuming that NHTSAâs (2017) Notice of Proposed Rulemaking to create FMVSS 150 evolves into an enforceable standard, all new vehicles will be equipped with a specific CV capability that can support a variety of transportation information and automation functions starting in the early 2020s. The profile of market penetration growth should then be reasonably predictable based on the rate of introduction of new vehicles and the retirement of old vehicles, although this could be accelerated if interest grows in the availability of after-market retrofit V2V communication systems. Deployment of the cooperative I2V/ V2I roadside infrastructure is less certain, since this will depend on deci- sions to be made by state, regional, and local government agencies that have not yet been convinced of the benefits they will derive from this. For- tunately, many of the benefits of CV technology for highway operations can be gained from use of the V2V data, without depending on the uncertain infrastructure investments. The uncertainties are much more significant for the AV technologies, which are not as mature for the higher levels of automation (SAE L3âL5). Those higher levels of automation are the ones that are likely to have the more profound implications for the future Interstates. The uncertainties cover several dimensions:
APPENDIX F 365 1. How capable will the automation technology be of performing the complete dynamic driving task within various operational design domains, and when will those capabilities first become available for use by the general public? 2. After each technology becomes available commercially, what will be the consumer interest and willingness to use it (either by pur- chasing it in their personal vehicles or using it through a shared mobility service)? How quickly will that interest grow, and how will that be modulated by the cost of the technology? Will that interest cool significantly after fatal crashes (killing innocent by- standers) caused by the automation technology are reported? 3. How much interest will there be in the United States in develop- ment of new greenfield (or brownfield) technology-focused cit- ies that could be designed from the start to have transportation infrastructure well suited to AV operations? If there is significant development along these lines, the AV technology could be used earlier and more widely than if it needs to wait until it is capable of handling the full complexity of traffic conditions on roads that must be shared with all other road users. None of these questions can be answered with any certainty today, which means that the prospects for widespread use of vehicles capable of higher levels of automation in the coming decades remain subject to a high degree of uncertainty. We could be looking at a future 50 years from now in which only specialized niche application vehicles are highly automated, or one in which a substantial fraction of all road vehicles are driven most of the time by automation systems. However, we can be certain that we will not be in a situation in which all road vehicles will be driven automaticallyâeven apart from the technological challenges, there is still a matter of consumer acceptance. Adoption of automation technology will be a voluntary consumer decision, and since there is a segment of the popula- tion that is actively hostile to automated driving, there will continue to be a market for new vehicles that are not highly automated. REGIONAL DIVERSITY OF IMPACTS Although the market for road vehicles is a national one, meaning that the same vehicles are available for use by travelers throughout the country, the roadway infrastructure is likely to be even more diverse across the country (across regions and between urban and rural land uses) than it is today. The future operations of the highway network will be more dependent on the cooperation between in-vehicle and infrastructure systems than they are now, given the growing importance of I2V and V2I CV systems to enable
366 NATIONAL COMMITMENT TO THE INTERSTATE HIGHWAY SYSTEM the vehicles and the roadway infrastructure to function as a well-integrated system. This means that it will be increasingly difficult to treat the National Highway System as a single consistent system unless extraordinary efforts are invested to promote nationwide consistency and uniformity. The forces working against national consistency are several and powerful: 1. Differing affinities of the populace for reliance on new vehicle technology and different levels of resources available to be early adopters (both supply- and demand-side effects). These differences are not only regional (sun belt versus rust belt or coastal versus inland) but also tied to the urban versus rural divide. 2. Different financial and human resources available to public agen- cies in different parts of the country to develop and operate I2V and V2I cooperative systems. These could represent differences between high-tax and low-tax states, prosperous versus depressed metropolitan regions, urban versus rural counties, or âself-helpâ jurisdictions (those that choose to tax themselves specifically for financing transportation improvements) versus those that are de- pendent on the largesse of their state or federal agencies. 3. These trends will be amplified by the growing digital divide within the population, which already tends to be correlated with these regional differences. The likely differences in regional economic growth (high-tech regions growing, rust-belt regions shrinking in population and economic activ- ity) are only going to exacerbate these forces at the national level. The one significant advantage that the less prosperous locations have is better availability and lower cost for new right-of-way that could be used for new transportation facilities or expansions of existing facilities. The de- velopment of new highway facilities that are designed specifically for use by more highly automated vehicles could be a major factor in leveling the field nationally in favor of the less prosperous or advanced locations that want to enjoy the transportation benefits from higher levels of road vehicle automation. PASSENGER TRAVEL DEMAND IMPLICATIONS OF CAV TECHNOLOGY The CAV technologies will have impacts on both the supply and de- mand sides of transportation, and those impacts will differ for passenger and freight movement. In this section and the subsequent sections, these
APPENDIX F 367 respective impacts are discussed in largely qualitative terms, since detailed quantitative estimates would be speculative at this early stage. The passenger travel demand implications of CAV technology are dis- cussed first, followed in subsequent sections by the freight travel demand and the combined supply-side implications. Note that the CAV technologies are likely to have diverse impacts on travel demand, with some tendencies toward increasing VMT and others toward decreasing VMT. With large uncertainties attached to both positive and negative tendencies, it becomes particularly challenging to predict the likely net impacts. The U.S. Depart- ment of Energy recently published a study to predict the net impacts on energy consumption based on two dimensions of future urban travelâthe level of automation and private personal versus shared vehicle usage (DOE 2017), and identified dramatically different impacts depending on which level of automation and which level of ridesharing become dominant. CAV and Related Technologies Reducing the Number of Vehicle Trips Telecommuting to Work at Home This is already a powerful trend, and it is not likely to weaken in the com- ing decades, but is only likely to strengthen. With the growth in employ- ment in information technology and related fields that depend on individual creative work on computers, this can be a very efficient way of accomplish- ing the work that is attractive to both employers and employees. It saves on costs of office space for employers and saves on commute time and expenses for employees, but it can make it more challenging for employers to assess work performance of their employees. It is also consistent with the growth of the âgig economy,â with people working as independent contractors rather than as regular employees, and often working fractions of weeks on an as-needed basis. Over the long term, this could lead to significant reduc- tions in the traditional peak-period commute traffic volumes on urban and suburban Interstate highways. Remote Work Centers, Closer to Homes This is an intermediate step between traditional office work patterns and telecommuting, allowing employees to do their work at shared worksites scattered throughout a region rather than at a central site. This may not reduce the number of trips as much as it reduces the length of the work trips that are taken to the remote work centers, but its impact on traffic volumes and patterns could still be substantial for urban and suburban Interstate highways in locations that are based on a knowledge economy.
368 NATIONAL COMMITMENT TO THE INTERSTATE HIGHWAY SYSTEM Teleconferencing and Virtual Reality Reducing Longer-Distance Trips to Meetings Teleconferencing is already having an influence on reducing long-distance business travel based on the potential to save significant time and money, but it is still a relatively narrow application because it is not yet a very convincing substitute for face-to-face contact for many purposes. There is a potential for great growth here, particularly as virtual reality or augmented reality can provide an impression of more direct personal contact among people who do not already know each other. This could become the pre- ferred mechanism for routine sales calls or for a variety of other meetings that currently require travel, which could in turn significantly reduce the number of business trips that are currently taken, with significant implica- tions for personal business travel demand on all categories of Interstate highways. Online Retail Reducing Shopping Trips Online retail is already having dramatic impacts on traditional brick-and- mortar retail, and that trend is likely to grow. The suburban shopping malls of the 1960s and 1970s are suffering serious decline, which is likely to ac- celerate, leading to significant changes in patterns of shopping travel. Trans- portation infrastructure that was built to serve those malls is likely to be underutilized unless the mall owners are able to redevelop their properties into destinations that are attractive for other purposes. This is an example of a technology-driven trend whose longer-term implications are difficult to discern, beyond the general observation that it is having a significant impact on shopping tripâmaking behavior, with likely implications for urban and suburban Interstate highway usage. Rideshare Matching and Transportation Network Companies Facilitating Ridesharing, Increasing Vehicle Occupancy This is perhaps the most widely discussed recent trend in transportation, particularly in its coupling with vehicle automation. There has been much overheated speculation about the end of private vehicle ownership, a poten- tially dramatic decline in the market for new motor vehicles, and the end of the need for parking spaces for vehicles that will be in continuous use. This speculation has been fueled by the entry of Uber, Google (Waymo), and many major vehicle manufacturers into the shared-use ride-sourcing market based on anticipated replacement of human drivers by automated driving systems. That type of future scenario would only be viable if Level 5 automation (or Level 4 automation with very few ODD limitations)
APPENDIX F 369 became technologically feasible and affordableâa prospect that is likely to come in the very distant future, if ever (for reasons to be explained in the section on effects at different planning horizons). The reductions in vehicle trips would only occur if large portions of the population were to become comfortable with sharing their rides (not just vehicles) with total strangers. There is little current evidence to indicate that this would be acceptable to the broader population when considering factors such as personal privacy, personal safety (especially for women or children traveling alone, and in the absence of an authority figure in the vehicle), and widely differing personal preferences regarding choices of entertainment and personal hygiene and behavior. There has also been some recent evidence from cities in which Uber and Lyft are serving large numbers of travelers to indicate that they are drawing more people away from conventional public transit than from private personal vehicles, leading to lower overall vehicle occupancy and worsened traffic congestion (Schaller Consulting 2017). Technology Changing the Character of Trips Apart from the trip reduction potential discussed in the section on CAV and related technologies reducing the number of vehicle trips, there is a po- tential for information technology to change the character of the trips that are taken even if the number of trips does not decline. There several ways in which future trip-making choices could change from the current norms. Improved Real-Time Traffic Condition and Route Guidance Information Leading to More Efficient Routing Travelers have been receiving real-time traffic condition information from mobile phone applications in recent years, helping them to avoid some of the worst traffic congestion problems. However, the information is still frequently inaccurate or obsolete, which limits its usefulness for decision making by travelers. In the coming years, the quality of this information should improve significantly, making it more attractive to a larger share of the traveling population. That should help reduce the current inefficiency in traveler responses to incidents, enabling them to find faster and less congested alternatives to the routes that have been affected by incidents. Although this will be good for travelers and traffic managers, it is not clear that it should have much impact on the design or construction of future Interstate highways, apart from the need to include provisions for the infor- mation infrastructure to support collection and dissemination of improved traffic condition data.
370 NATIONAL COMMITMENT TO THE INTERSTATE HIGHWAY SYSTEM Parking Information Reducing Wasted Travel Seeking Parking A significant proportion of urban driving is currently wasted mileage by drivers seeking parking spaces (Shoup 2007). Improved parking informa- tion collection and dissemination should help reduce that wasted mileage and the associated impacts on urban traffic congestion, pollution, and energy consumption. However, it is unlikely to have much impact on the Interstate highways of the future. Reduced Importance of Traditional Shopping Malls and Office Parks as High-Volume Destinations Patterns of urban land use and activity are likely to change in the coming decades as a consequence of some of the changes in work patterns and increased reliance on the Internet. The impacts on shopping malls were already discussed, but office parks are also potentially subject to decline if more people telecommute. This means that some of the current edge-city activity centers could decline in importance, especially if they lack other reasons for being. Highway infrastructure that was built to serve these sites could become obsolescent as activity patterns change unless the owners of these sites are agile in finding or creating other uses for them. If Traditional Commute Work Trip Patterns Decline by Enough, Special-Event Travel Could Become the Defining Case to Determine Capacity Needs The current urban and suburban Interstate highway network has in large part been scaled to serve the morning and evening commuter peak travel demand, which is the heaviest travel demand experienced in most loca- tions in the United States. If the changes in information technology cause work trip patterns to change significantly, the peak travel demands could be defined differently, based on considerations such as sporting events, festivals, holiday periods (skiing weekends), or emergency evacuations in locations that are vulnerable to weather emergencies. This could put pres- sure on planners to rethink how they determine the capacity needs for their Interstate highway networks, which may find themselves overbuilt in some locations but needing expansion in others. Technology Increasing the Number and Length of Vehicle Trips Just as information technology has the potential to reduce travel demand, it also has the potential to increase demand in other ways. With the large uncertainties on estimates of both increases and decreases in travel demand,
APPENDIX F 371 it becomes particularly challenging to estimate what the net changes in demand are likely to be. Empty Backhaul Trips for Repositioning Shared-Use Vehicles to Make Their Next Trip The concept of shared-use vehicles (car sharing) currently depends on the vehicles being returned to a fixed location where they can be accessed by the next user or (in the future) on a highly automated driving system tak- ing the vehicle directly to its next userâs origin. The attraction of the latter concept is increased convenience for the users and higher utilization of the vehicles, but it involves empty backhaul mileage for the trip between the first userâs destination and the second userâs origin. Several case stud- ies have included estimates of the amount of extra travel that would be required, which depends heavily on the local travel patterns and the size of the shared-use fleet. For example, Viegas and Martinez (2017) showed that when they simulated the use of 8- and 16-passenger shared taxi services to provide the feeder service to rail transit and all other passenger trips in the Lisbon urban region, those vehicles would be operating 20 percent of the time with no passengers onboard while they deadhead from one passenger- carrying trip to the next. In a related study for Lisbon, Viegas and Martinez (2015) showed that the overall weekly VMT would increase by 6 percent from todayâs base- line if their shared taxis were deployed in combination with high-capacity metro service or by 22 percent if all trips were to be served by the shared taxis in the absence of high-volume public transit service. Fagnant and colleagues (2015) simulated shared automated vehicle services for a 12 Ã 24-mile region in Austin, Texas, and found that empty backhaul travel added 8 percent to the VMT for the area, with an average wait time of 1 minute. These studies have been for relatively short-distance urban driving applications, but it is not clear how viable the shared-use concept will be for the longer trips that would be more likely to use urban and suburban Interstate highways. High Automation of Driving Allowing âDriversâ to Make Productive Use of Travel Time One of the major potential benefits to users of highly automated vehicles is the notion of the electronic chauffeur that could take over most or all of the driving under some road and traffic conditions so that the user could use the traveling time to do other things (work, play, or sleep). This severely reduces the effective cost of that travel time to the user, which could encour- age people to take more and longer trips, including choosing to live farther
372 NATIONAL COMMITMENT TO THE INTERSTATE HIGHWAY SYSTEM from their workplaces. This triggers one of the major potential concerns about highly automated drivingâthat it will release significant latent de- mand and may induce new travel demand by encouraging urban sprawl. This should be a major consideration for the long-term planning of future Interstate highway developments, which may need to accommodate such demand unless policies are implemented to actively discourage it. High Automation of Driving Enabling More Travel by People Who Currently Cannot Drive or Are Intimidated by Highway Driving Highly automated driving systems offer the prospect of restored mobility to senior citizens who are no longer able to drive themselves, as well as relieving parents of the burden of chauffeuring their children everywhere they need (or want) to go. These potential increases in mobility of the nondriving population also raise the prospect of releasing latent demand and thereby increasing the overall volume of travel, since trips that are cur- rently too costly or inconvenient could become significantly less costly or inconvenient. This could lead to a growth of up to 14 percent in VMT, as projected in recent research that assumes nondrivers would travel as much as their driving counterparts in the same age and gender categories (Harper et al. 2016). The implications may be more significant for short-distance trips on local streets than for long-distance Interstate highway trips. High Automation Advances to the Level of Enabling Automated Taxi Services on Freeways Could Lead to Increased Demand for Urban Freeway Traffic Most of the current consideration of automated taxi-like services by major automotive companies and technology companies such as Google/Waymo and Uber has been focused on low-speed, short-range, urban applications. However, when the automation technology improves to the level that it can be entrusted with freeway driving, there could be an impact on urban and suburban Interstate highway traffic, based on the addition of empty back- haul driving to reach the next traveler. If the taxi service is also based on shared occupancy of the vehicles, the higher occupancy could compensate by eliminating some vehicle trips. High Automation of Driving Could Encourage Remote Parking of Personal Vehicles on Low-Cost Land in Peripheral Locations One of the long-term visions of the highest levels of automated driving has been eliminating parking from high-density urban cores and relying on the
APPENDIX F 373 automated driving system to drive the empty vehicle back and forth be- tween the userâs destination in the urban core and a remote parking facility on cheaper land in the outskirts. This would create extra empty mileage for the round trips between the travel destination in high-cost locations and the remote parking sites, which is likely to have significant adverse implications for traffic volume, energy, and the environment. The long-term implications need to be considered seriously for urban and suburban Interstate highways that would provide access to the remote parking sites, but this is unlikely to become technologically feasible for several more decades (and it will take further decades for the population of suitably equipped vehicles to become dominant). If the Cost of Travel Declines Dramatically, Consider Growth of New Types of Zero-Occupancy Trips for Nontravel Purposes We normally think about the demand for travel being a derived demand, based on the need to move people or freight from one location to another. However, vehicles could be driven to serve other purposes if their cost of operation were low enough. This is where the concept of the mobile bill- board could come into play, a vehicle that is just driven throughout the roadway network so that it can be seen by the people in other vehicles or along the roadside. It would represent a new form of demand for VMT with zero occupancy, which raises broader policy questions about whether it should be condoned. It has the potential to add to the demand on major Interstate freeways that have high enough traffic volumes to be interesting to advertisers. FREIGHT TRAVEL DEMAND IMPLICATIONS OF CAV AND RELATED TECHNOLOGIES The implications of changes in the demand for goods movement are likely to be at least as significant for the future of the Interstate Highway System as the changes in movement of people. The dominant factors in goods movement demand are much larger economic forces, such as the levels of international trade with different trading partners (especially Canada and Mexico), the type of goods involved in that trade, and the overall health of the different sectors of the U.S. economy and of its regions. The highway system is only directly affected by the demand for goods movement by truck, but that is in turn affected by truckingâs service and price competi- tiveness with other modes for the goods that could potentially be moved by multiple modes. This is where the influence of CAV enters the mix, because it can improve both service and price competitiveness of trucking.
374 NATIONAL COMMITMENT TO THE INTERSTATE HIGHWAY SYSTEM The CAV influence on service improvements includes â¢ Better real-time traffic and weather information enables truck op- erators to choose better routes and dynamically change routes to avoid delays, reducing delays and improving delivery time reliability. â¢ Operation of trucks using cooperative ACC and platooning in- creases the capacity and smooths traffic on congested truck cor- ridors, reducing delays and improving reliability of delivery times. â¢ Use of Level 3 and Level 4 automation to take over tedious driving tasks makes truck driving a more attractive occupation, helping to relieve the current shortage of truck drivers. â¢ Level 3 and Level 4 automation eventually enables modification of driver hours-of-service rules so that drivers can work longer shifts without fatigue and trucks can complete longer-haul delivery runs sooner, increasing their competitiveness with air for higher-value shipments. â¢ When Level 4 automation matures to the point that a truck could drive the entire length of an Interstate highway trip without a driver onboard, the trucks could be driven continuously without regard to hours-of-service limitations, limited only by the distance between necessary refueling stops. These service improvements can also help reduce truck operating costs, making it possible for trucking to become more price competitive with rail, and perhaps increasing the length of haul at which trucking is seen as the best modal alternative. Additional opportunities for CAV to improve the price competitiveness of trucking include: â¢ CACC and platooning enable trucks to drive closer together, saving 10 percent to 15 percent of their fuel consumption costs through reduced aerodynamic drag. â¢ Efficiency and traffic flow improvements already cited above in the service category provide further fuel consumption reductions, saving additional operating costs. â¢ In the longer term, when CAV technology enables the following trucks in a platoon to be driven without a driver onboard, the driver labor costs could be eliminated, leading to a significant op- erating cost saving. However, this concept has further implications for the highway infrastructure because it would probably require the development of staging areas at major freeway entry and exit points (analogous to railroad marshaling yards) where the platoons would be assembled and disassembled and where control would be
APPENDIX F 375 transferred between drivers (who would still drive the local pickup and delivery portions of the trips) and the automation system. â¢ In the even longer term, when CAV technology enables any truck on an Interstate highway to drive without a driver, the labor costs for truck operations could be further reduced. The implications of these potentially significant changes in use of trucks on Interstate highways will not be uniform across the Interstate network, but are likely to differ between long-haul operations on rural highways and short-haul operations on urban highways (which are more likely to be dominated by port drayage and pickup and delivery operations). The long-haul operations will be more influenced by the drag reduction and driver labor-saving opportunities, whereas the short-haul operations will be most affected by the congestion reduction opportunities. The timing for realizing these benefits is also likely to be influenced by the availability of resources for development of dedicated truck lanes where the trucks can take maximum advantage of these opportunities to improve their opera- tions (especially for the higher levels of automation). Future Interstate high- way planning should take serious account of how to design, operate, and finance such dedicated trucking facilities within the Interstate right-of-way. Other advances in information technology not directly classifiable as CAV are also going to have important effects on the demand for goods movement on the future Interstate Highway System. Further growth of online shopping will continue to change consumer product distribution pat- terns, with more focus on direct home delivery rather than delivery to retail stores. This could affect decisions about locations of warehouses and both geographical and temporal distributions of deliveries, especially on urban and suburban freeways. Furthermore, the growth of local three-dimensional (3-D) printing for manufacturing could shift freight demand away from fabricated products and more toward the bulk raw materials used by the 3-D printers. It is not clear yet whether that may favor rail shipment of the bulk materials over truck shipment of the fabricated products. SUPPLY-SIDE IMPLICATIONS OF CAV TECHNOLOGIES FOR THE INTERSTATE HIGHWAY SYSTEM The most direct impacts of the CAV technologies on the future Interstate Highway System will be on the supply side rather than on the demand side. These technologies will produce significant changes in the characteristics of vehicle traffic, with important implications for safety and efficiency of vehicle movements. The consequences for design and operation of the highway system are not yet entirely clear, but will need careful study. The diversity of changes is at least as broad as the diversity of CAV alternatives that are likely to be implemented.
376 NATIONAL COMMITMENT TO THE INTERSTATE HIGHWAY SYSTEM Collision Warning and Avoidance Systems These systems could potentially reduce current crash rates by about half. Since crashes have been estimated to cause about 28 percent of the con- gestion on urban freeways (Varaiya 2005), this could potentially reduce nonrecurrent highway congestion by about half that amount, leading to reduced delays and improvements in travel time reliability. Improved Traffic Management and Incident and Weather Management These applications, which are enabled by CV technology, could further re- duce congestion impacts of incidents and weather and facilitate emergency evacuations of low-lying coastal areas threatened by sea level rise and extreme weather events. Traffic Management Strategies Using Variable Speed Limits (VSLs) and Coordinated Ramp Metering (CRM) These strategies can be implemented based on V2I collection of traffic probe information, with recommendations provided to drivers using roadside variable message signs, or they could depend on I2V information to provide in-vehicle displays to drivers or direct access to control the maximum set speeds of ACC systems. As the level of information and control increases, they should have increased impacts on improving the effective capacity of freeway bottlenecks, partially relieving some choke points, and improving incident response (Lu et al. 2015). Integrated Corridor Management (ICM) Increased use of CV technology to collect and disseminate real-time traffic information can help managers of freeway, arterial, and transit networks to integrate their operations to provide more effective utilization of the capac- ity available in urban and suburban corridors, especially when responding to incidents that disrupt some of these networks. This will reduce delays and improve trip time reliability for a wide range of travelers and vehicles. Advance Reservations for Highway Trips CV technology offers the potential for travelers (private personal travelers or truck drivers) to request advance reservations for trips on congested highways, guaranteeing them preferential access at their reserved times. This could help significantly in spreading the peak congestion periods, encouraging travelers who do not reserve as early or not at all to travel
APPENDIX F 377 at times away from the peak. This could of course also be coupled with congestion charging via CV technology, adjusting the prices for travel based on time of day or current conditions, further encouraging peak spreading. Extended Electronic Toll and Traffic Management Electronic toll collection technology is already in widespread use and is dis- placing traditional cash-based toll collection for a variety of reasons. This trend should accelerate, and the technology should enable, more advanced applications of dynamic electronic road pricing based on real-time traffic condition information. This enables both traffic managers and travelers to make real-time decisions based on the most up-to-date information about current traffic conditions and problems. Right-Sized Public Shared-Use Automated Transit Vehicles The future generations of public transit service are unlikely to look like todayâs 40-foot buses, which have not been very successful at attracting riders in all but a handful of cities. In the future, when Level 4 automation technology makes it possible for a bus to operate without a driver on its own urban busways or on bus lanes on the Interstate System, a more flex- ible type of transit service will be possible, with vehicles sized to the antici- pated volume of traffic demand because the driver labor cost is no longer the driving factor. Applying CACC or platooning technology to the transit vehicles in the bus lane can help to provide higher passenger throughput in locations where that is a potential concern. Such bus lanes could become important elements of the future urban Interstate highway network. Automated Truck Platoons This is one of the initial CAV operational improvements that could be achieved within the next 5 years, based on technology that is already in advanced prototype testing (Dutch Ministry of Infrastructure and the Envi- ronment 2016; Tsugawa et al. 2016). It can provide significant lane capacity and traffic flow stability improvements while also saving operating costs for fleet operators, potentially inducing more truck demand. The impacts on operations depend strongly on market penetration, and so it will take some time for the percentage of equipped trucks to grow to the level that will have a significant impact on Interstate corridors with high volumes of truck traffic. The more advanced version of this technology, which could gain larger operating cost savings by operating the following trucks without drivers, would be able to enter the market for public use more quickly if dedicated
378 NATIONAL COMMITMENT TO THE INTERSTATE HIGHWAY SYSTEM truck lanes or corridors were available on key Interstates to segregate the trucks from the hazards posed by recklessly driven cars and motorcycles. This is an alternative that should be explored on the infrastructure side now so that it could be implementable in 10 years, when the driverless platoon follower technology could become feasible for use within such a restricted ODD. Automated Urban Freight Distribution CAV technology will produce a variety of innovations in urban freight distribution, especially with the potential for smaller and more specialized delivery vehicles that could operate without drivers under certain ODD restrictions in future decades. These will probably have more of an impact on local urban traffic than on Interstate highway operations, although there could be some impacts on urban Interstate highways near goods distribu- tion hubs. Cooperative Adaptive Cruise Control and Automated Merge Coordination These systems combine CV and AV technology to increase the effective density of highway traffic without producing traffic instabilities, thereby significantly increasing throughput at high market penetration (up to a factor of 2) (Shladover et al. 2012). In this way, the existing physical infra- structure of an Interstate freeway could potentially accommodate twice as many vehicles per hour as it does today, without reductions in speed or flow breakdowns (no stop-and-go disturbances). This can be accomplished with Level 1 automation (speed control only) and V2V communication, and so it does not depend on dramatic technological advances, but only on increased market penetration of vehicles with the needed capabilities. Level 4 Automation on Freeways This level of automated driving within the ODD of limited-access freeways could support dramatic improvements in capacity and congestion reduc- tion, especially if lane(s) could be dedicated for their use (maybe up to a factor of 3 in capacity per lane compared to no automation) (Michael et al. 1998; National Automated Highway Systems Consortium 1997). This is a situation in which the combination of the in-vehicle automation technology and the dedicated physical infrastructure can produce significantly more benefit than either one or the other in isolation. These substantial capacity increases on freeways could be the safety valve that reduces the need for much additional civil infrastructure, but
APPENDIX F 379 better integration with local arterial traffic operations will be necessary as part of the implementation to avoid simply shifting the bottlenecks to the arterials and freeway entrances and exits. The benefits to the freeway network operations should be sufficiently large to justify the relatively modest investments that are likely to be required for the improved arterial coordination. The cost of providing connectivity and system management capabilities on the roadway infrastructure should be a minuscule fraction of the costs of the civil infrastructure construction and right-of-way, but the operational and maintenance costs will be a more significant fraction of the deployment costs, so they need to be factored into the financing model explicitly. Main- tenance cost considerations will have to include better visibility of pavement markings and signage to help computer vision systems on vehicles, and also management of pavement wear if vehicles all follow the same paths very ac- curately (this can be resolved by deliberately introducing random misalign- ments into the vehicle steering guidance algorithms). The current Interstate financial support model will need to be reconsidered, since it is based on the more traditional highway technologies, in which the large majority of the life-cycle cost of the system is in the original capital construction costs. It will be important to avoid creating disincentives for states to invest in the most cost-effective highway alternatives because the funding formula favors support for capital costs over operating and maintenance costs. Cybersecurity introduces a new set of issues for highway designers and operators to consider. There is a significant risk of transportation system disruption if vehicles are hacked, and there is a need for (probably standardized approaches to) protection of the traffic management systemsâ information infrastructure (which in turn affects infrastructure operations and maintenance costs and staffing). EFFECTS AT DIFFERENT PLANNING HORIZONS The previous sections of this appendix have discussed the types of impacts that information technology is likely to have on the future Interstate High- way System in general terms, which is the easier part of this look ahead. The harder part is discerning how long it is likely to take for the different impacts to occur, based on the uncertainties about the pace of technology advance and the rate of user acceptance of the technology after it has been developed. Further uncertainties arise based on future changes in economic activity, both national and international, and in the demographics of the U.S. population. The effects under consideration here are related to the overall study goals of serving network traffic flows more efficiently and exploiting in- novation and advances in technologies to improve system safety, resilience,
380 NATIONAL COMMITMENT TO THE INTERSTATE HIGHWAY SYSTEM management, and operations. This is a broad set of potential impacts, which can at best be discussed in semiquantitative terms, rather than getting into highly refined numerical exercises. Following the guidance of the com- mittee, the impacts are considered based on planning horizons of 10, 20, and 50 years, with increasing levels of uncertainty as the horizon recedes. 10-Year Horizon The 10-year horizon forecast is dominated by the inertia in the current transportation system, which cannot change rapidly on either the vehicle or infrastructure side. Many of todayâs unequipped vehicles will still be on the road at that time, and the vehicles with more advanced capabilities will still be a small fraction of the new vehicles purchased each year, so they are likely to remain a small fraction of the overall vehicle population. Information technology modifications will be progressing gradually on the infrastructure side if funding is available to pay for them, but physical mod- ifications to the roadway infrastructure are not likely to have progressed far, given the length of the lead times involved for the full environmental review and construction processes. â¢ Expect significant connectivity across the vehicle fleet, with favor- able impacts on safety and traveler information and trip planning, based on the assumption that the NHTSA mandate to deploy V2V communication systems for safety proceeds as planned. If that mandate is dropped or the DSRC spectrum is lost, this impor- tant enabler of many enhanced functions will at best be delayed significantly. â¢ Highway operators should be deploying infrastructure for I2V and V2I connectivity and transportation network management to support the most efficient operations. This will probably be primar- ily in the states and regions that are most advanced and have the most resources, but it is not likely to be evenly distributed around the country. The extent of these deployments will depend heavily on the policy and funding decisions that are taken by the federal government during the coming decade. â¢ Limited applications of partial freeway automation will be in place based on sales of Level 1 and Level 2 systems to consumers. The Level 1 ACC systems could be quite widespread, but it is not clear how many of them will be combined with the V2V communication systems to produce the enhanced CACC capabilities. The Level 2 automation systems will probably still be at low market penetra- tions, not yet having significant impacts on traffic conditions, since customers do not receive that large a benefit for the additional cost.
APPENDIX F 381 The net effects on freeway operations will probably be too small to measure throughout most of the country. â¢ Truck CACC and platooning systems will be used by some major truck fleets on major freight corridors and will be exposing the public to the experience of sharing the road with partial automa- tion systems. They will probably not be in wide enough use to have a significant impact on traffic conditions. In summary, the changes from today will be relatively modest, and there is a low uncertainty on that prediction because of the large inertia in both the vehicle and infrastructure systems that impedes rapid changes. 20-Year Horizon The 20-year forecast has a much larger uncertainty than the 10-year fore- cast, especially with regard to the vehicle automation technology and its impacts. However, it appears highly likely that vehicle connectivity of one type or another should be virtually ubiquitous by then, providing comprehensive information to travelers and transportation system opera- tors to assist them to make better decisions. The fate of V2V connectivity for collision warning and cooperative automation is not as certain, since this depends to a considerable extent on whether the NHTSA rulemaking requiring DSRC broadcasts by all new vehicles is actually implemented. Even if this rulemaking does not go forward, there is a reasonable chance that other wireless technologies, based on 5G cellular, could serve a similar purpose, but with some delay in implementation. With regard to the vehicle automation technologies, the topics that are most important for the future operations of the Interstate Highway System are: â¢ Truck platooning should be commonly available, saving energy and emissions and improving traffic flow, and its economic benefits in terms of labor cost savings could be greatly increased if dedicated truck lanes are developed in major Interstate freight corridors. â¢ CACC and platooning of passenger cars and transit vehicles could produce significant operational improvements for urban Interstates (reducing congestion, energy use, and emissions), especially if dedi- cated lanes are provided for their use. â¢ Level 4 cooperative automation of vehicles for operation on well- protected dedicated lanes within Interstate highways should be feasible, and building such lanes will provide a significant stimulus to the development and use of these vehicles. These dedicated lanes for cooperative highly automated vehicles could offer substantially
382 NATIONAL COMMITMENT TO THE INTERSTATE HIGHWAY SYSTEM higher capacity, safety, and possibly speed than normal highway driving, producing substantial impacts in the locations where they are implemented. If dedicated lanes are not provided, the Level 4 automation system capabilities and impacts will be much more limited and it will take significantly longer for them to come to market because of the technical challenges of ensuring their safety in mixed traffic. There is a moderately high level of uncertainty about how rapidly these changes will occur and how large their net impacts will be. 50-Year Horizon The 50-year forecast is fraught with extremely large uncertainties because of the large variety of influencing factors that could change in dramatic ways. The overall economy and society could change in ways that we can- not imagine today, and technologies that we cannot envision today could become reality. The 50-year time frame could be adequate to resolve the daunting technological challenges to Level 5 automation of road vehicles, and it could also produce other technological changes that might dramati- cally reduce the need for road travel (virtual reality encounters substituting for live entertainment and/or business meetings, telecommuting dominating work environments in many occupations, alternative modes of travel such as the Hyperloop becoming highly competitive, and so forth). Absent these kinds of revolutionary developments, the influences of CAV technologies on the Interstate Highway System in 50 years are likely to include the following: â¢ High levels of (cooperative) automation are likely to be in wide- spread use for all classes of vehicles on the Interstate highway network, providing significant operational and safety improve- ments. When the market penetration of the more highly automated vehicles reaches a suitable threshold, it becomes politically easier to segregate these vehicles from the manually driven vehicles, leading to large increases in throughput for the lanes with the highly auto- mated vehicles. The energy consumption and emissions from these vehicles should be reduced from current levels based on smoother speed profiles and reduced aerodynamic drag from shorter fol- lowing distances. However, if the infrastructure investments for segregated roadway infrastructure are not made, the throughput and efficiency gains will be reduced significantly. â¢ Regardless of the rate of progress with highly automated vehicle technology, the vehicles driving on the Interstate highways will
APPENDIX F 383 continue to include some conventional manually driven vehicles and vehicles with lower levels of driving automation capabilities. These will be used by people who cannot afford the newest ve- hicles, people who prefer to drive their legacy vehicles, and people who are opposed to highly automated driving for various personal reasons. The percentages of these vehicles are likely to vary widely among urban regions and between urban and rural highways even after highly automated freeway driving is generally available on new vehicles (just as the percentages of vehicles with sunroofs or ACC vary widely today). â¢ Widespread use of automation could be influencing locational de- cisions and travel patterns, with potential for both positive and negative societal impacts. If the low cost and ease and convenience of automated road travel releases latent travel demand, this could lead to significant increases in the volume of road traffic. Over the long term, if this leads to changes in land use patterns and induces people to travel longer distances to satisfy their regular needs, it could also induce new and longer trips, producing even larger increases in the volume of road traffic. Policy makers will need to confront these challenges to determine what compensatory measures may be needed to discourage excessive growth in travel, with its concomitant energy and environmental costs. Concluding Note on Technological Challenges These projections are more conservative than most published and widely cited predictions, based on concerns about several severe technological challenges that need to be overcome before it will be possible for software to drive road vehicles at least as safely as human drivers can. This means that the road transportation system will continue to be populated by a mix- ture of vehicles with widely varying levels of automation for the foreseeable future. Manually driven vehicles will continue to be part of the mix, along with vehicles using the lower levels of automation to enhance the traveling experience, even after more highly automated vehicles become available for public use. Automation systems at Level 3 and above produce fundamental changes in the driving process, which create even larger challenges for technology than they do for regulations. For the first time, technological elements are taking the primary responsibility for ensuring the safety of the vehicle oc- cupants and other road users away from the human driver. At Level 3, this responsibility is taken temporarily and may be returned to the human driver on short notice (several seconds), but at Level 4 it is taken over for a sus- tained period (as long as the vehicle remains within its specified ODD), and
384 NATIONAL COMMITMENT TO THE INTERSTATE HIGHWAY SYSTEM at Level 5 it may be taken completely. Although no explicit safety standards have been specified yet, it is not unreasonable to expect the automated driv- ing system to maintain at least the level of safety of average human drivers today (some observers contend that it is more likely to require 10 times the safety of average human drivers in order to be socially acceptable). As explained in Shladover (2014), for road travel in the United States today, this represents a mean time between fatal crashes of more than 3 million vehicle hours of driving (representing 375 years of continuous driving 24 hours per day, 7 days per week) and a mean time between injury crashes of about 65,000 vehicle hours of driving (representing more than 7 years of continuous 24/7 driving). Those numbers are similar for most industrial- ized countries (within a factor of 2 above or below), and that poses severe challenges for a software-intensive system that must operate in a highly dynamic and stochastic environment. Several serious technological challenges need to be conquered before automated driving systems will be able to safely operate without constant human supervision. These are summarized here, in order of increasing dif- ficulty, and are described in more detail in Shladover (2014) and Shladover and Bishop (2015). 1. Providing the automated driving system with comprehensive fault detection, identification, and accommodation capabilities so that it can immediately diagnose its own malfunctions and switch to a fallback mode of operation that can maintain safety even if it needs to sacrifice performance (such as significantly reducing speed and/or parking the vehicle on the shoulder of the road). This re- quires redundancy of hardware and software functionality, which is bound to increase costs of development and implementation. 2. Ensuring sufficient cybersecurity protection to repel the large ma- jority of cyberattacks. This is already becoming a challenge for modern nonautomated vehicles because of their dependence on electromechanical actuation (engine, brake, and steering control) and in-vehicle networks, but the temptation for attackers is likely to be greater for highly automated vehicles, whose occupants are less likely to notice and respond quickly to an attack (because they are not doing the driving). 3. Developing comprehensive environment perception capabilities that can reliably identify, track, and discriminate between benign and hazardous objects in the path of the vehicle under the full range of environmental conditions in which the vehicle is intended to operate (weather and lighting conditions). Essentially all hazard- ous objects must be recognized, even if they are difficult to perceive from a long enough range to enable the system or the driver to
APPENDIX F 385 take corrective action (potholes, rocks, or bricks in the path of the vehicleâs tires, and so forth). At the same time, the system must be intelligent enough to ignore nearly all benign objects (paper bags, balloons, newspapers, and so forth) even if they are highly visible so that the vehicle does not take spurious avoidance maneuvers, which will disconcert the vehicle occupants and could potentially cause new crashes. This is likely to require the fusion of data from multiple sensors that are based on different phenomenology and are not vulnerable to common-mode faults, which has cost implications. 4. Resolving questions of ârobot ethicsâ sufficiently to enable the system software to make âlife or deathâ decisions affecting the safety of all road users. Even if the environment perception soft- ware obtains âperfectâ knowledge of the environment surrounding the vehicle, it will still be confronted with questions about which target objects to hit when a crash is unavoidable, and the complex- ity of those decisions is magnified when the knowledge of those objects is confounded by uncertainties. This could be one of the first instances in which software is entrusted with the authority to make life-or-death judgments about multiple people, yet there are no established ground rules for making such judgments ethically or even for managing the design process. Similarly, ethical conun- drums arise when practical considerations of driving in imperfect traffic conditions conflict with strict interpretations of traffic law regarding speeds, crossing lane boundaries, and so forth. More broadly, designers of automation systems need to be made aware that they are applying ethical considerations in their work even if they are not conscious of that, and it is much better to be making those value judgments consciously rather than unconsciously. 5. Designing a software-intensive system for a very high level of safety, so that the rate of errors in the system requirements, speci- fications, and coding is sufficiently low that the system will be no less safe than human driving. This is the most daunting of all the technological challenges because there is no existing technology that can support the design, development, verification, or valida- tion of software of the level of complexity that will be needed for automated driving systems. Formal methods have been applied to much simpler software examples, but their complexity is such that they do not scale well to software of this complexity. The current methods of software verification and validation are very costly and labor intensive, even for applications that are much less compli- cated than automated driving (e.g., aircraft autopilots), and even those depend on a priori assumptions about the completeness of the software specifications, which cannot be ensured in this case.
386 NATIONAL COMMITMENT TO THE INTERSTATE HIGHWAY SYSTEM ACKNOWLEDGMENT Sections of this report are derived in part from an article, âConnected and Automated Vehicle Systems: Introduction and Overviewâ by Steven E. Shladover, to be published in the Journal of Intelligent Transportation Systems, Date TBD, Copyright Taylor & Francis, available online at http:// dx.doi.org/10.1080/15472450.2017.1336053. REFERENCES Abbreviations DOE U.S. Department of Energy NHTSA National Highway Traffic Safety Administration Anderson, J. M., N. Kalra, K. D. Stanley, P. Sorensen, C. Samaras, and O. A. Oluwatola. 2016. Autonomous Vehicle Technology: A Guide for Policymakers. RR-443-2-RC, 2016. RAND Corporation, Santa Monica, Calif. Audi. 2017. Safety, Comfort, and Efficiency: The Assistance Systems of Audi. http://www.audi. com/en/innovation/piloteddriving/assistance_systems.html. Bel Geddes, N. 1940. Magic Motorways. Random House, New York. Bender, J. G. 1991. An Overview of Systems Studies of Automated Highway Systems. IEEE Transactions on Vehicular Technology, Vol. 40, No. 1, pp. 82â99. Blanco, M., J. Atwood, H. M. Vasquez, T. E. Trimble, V. L. Fitchett, J. Radlbeck, G. M. Fitch, S. M. Russell, C. A. Green, B. Cullinane, and J. F. Morgan. 2015. Human Factors Evaluation of Level 2 and Level 3 Automated Driving Concepts. Report No. DOT HS 812 182. NHTSA, Washington, D.C. Dickmanns, E. D. 2002. Vision for Ground Vehicles: History and Prospects. International Journal of Vehicle Autonomous Systems, Vol. 1, No. 1, pp. 1â44. DOE. 2017. The Transforming Mobility Ecosystem: Enabling an Energy Efficient Future. https://energy.gov/eere/vehicles/downloads/ transforming-mobility-ecosystem-report. Dutch Ministry of Infrastructure and the Environment. 2016. European Truck Platooning Challenge 2016: Creating Next Generation Mobility: Lessons Learnt. http://www.eutruck platooning.com. Fagnant, D. J., K. M. Kockelman, and P. Bansal. 2015. Operations of Shared Autonomous Vehicle Fleet for Austin, Texas, Market. Transportation Research Record: Journal of the Transportation Research Board, No. 2536, pp. 98â106. https://doi.org/10.3141/2536-12. Fenton, R. E., and R. J. Mayhan. 1991. Automated Highway Studies at the Ohio State Uni- versity: An Overview. IEEE Transactions on Vehicular Technology, Vol. 40, No. 1, pp. 100â113. Glathe, H.-P. 1994. The PROMETHEUS Program: A Cooperative Effort of the European Automotive Manufacturers. Presented at SAE Brazil 94 Conference, Society of Automo- tive Engineers, Sao Paulo. Harper, C., C. T. Hendrickson, S. Mangones, and C. Samaras. 2016. Estimating Potential In- creases in Travel with Autonomous Vehicles for the Non-Driving, Elderly and People with Travel-Restrictive Medical Conditions. Transportation Research Part C, Vol. 72, pp. 1â9. Jutila, S. T., and J. M. Jutila. 1986. Diffusion of Innovation in American Automobile Industry. Presented at the Advanced Summer Institute in Regional Science, University of Umea, Sweden.
APPENDIX F 387 Lu, X.-Y., S. E. Shladover, I. Jawad, R. Jagannathan, and T. Phillips. 2015. Novel Algorithm for Variable Speed Limits and Advisories for a Freeway Corridor with Multiple Bottle- necks. Transportation Research Record: Journal of the Transportation Research Board, No. 2489, pp. 86â96. Markoff, J. 2017. Robot Cars Canât Count on Us in an Emergency. New York Times, June 7. https://www.nytimes.com/2017/06/07/technology/google-self-driving-cars-handoff- problem.html?hpw&rref=automobiles&action=click&pgtype=Homepage&module=w ell-region®ion=bottom-well&WT.nav=bottom-well&_r=0. Michael, J. B., D. N. Godbole, J. Lygeros, and R. Sengupta. 1998. Capacity Analysis of Traffic Flow over a Single-Lane Automated Highway System. Intelligent Transportation Systems Journal, Vol. 4, No. 1â2, pp. 49â80. National Automated Highway Systems Consortium. 1997. Automated Highway System (AHS) Milestone 2 Report: Task C2, Downselect System Configurations and Workshop #3. Troy, Mich. https://path.berkeley.edu/sites/default/files/ahs-milestone_2_report_task-c21. pdf. NHTSA. 2017. Federal Motor Vehicle Safety Standards; V2V Communications: Notice of Proposed Rulemaking. Federal Register, Vol. 82, No. 8, pp. 3854â4019. Rillings, J. H. 1997. Automated Highways. Scientific American, Vol. 277, No. 4, pp. 80â85. SAE International. 2016. Taxonomy and Definitions for Terms Related to Driving Automation Systems for On-Road Motor Vehicles. Surface Vehicle Recommended Practice J3016. Schaller Consulting. 2017. UNSUSTAINABLE? The Growth of App-Based Ride Services and Traffic, Travel and the Future of New York City. http://schallerconsult.com/rideservices/ unsustainable.pdf. Shladover, S. E. 1990. Roadway Automation TechnologyâResearch Needs. Transportation Research Record, No. 1283, pp. 158â167. Shladover, S. E. 2000. Bus Rapid Transit and AutomationâOpportunities for Synergy. In Proceedings of Seventh World Congress on Intelligent Transport Systems, Turin, Italy. http://onlinepubs.trb.org/onlinepubs/archive/conferences/VHA-BRT/Bus_Rapid_Transit_ and_Automation--Opportunities_for_Synergy.pdf. Shladover, S. E. 2001. Opportunities in Truck Automation. In Proceedings of Eighth World Congress on Intelligent Transport Systems, Sydney, Australia, Paper No. ITS00155. Shladover, S. E. 2014. Technical Challenges for Fully Automated Driving Systems. 21st World Congress on Intelligent Transport Systems, Detroit, MI. Shladover, S. E. 2016. The Truth About âSelf-Drivingâ Cars. Scientific American, Vol. 314, No. 6, pp. 52â57. Shladover, S. E., and R. Bishop. 2015. Road Transport Automation as a PublicâPrivate Enter- prise. In Conference Proceedings 52: Towards Road Transport Automation: Opportu- nities in PublicâPrivate Collaboration. Summary of the Third EU-U.S. Transportation Research Symposium, Transportation Research Board, Washington, D.C., pp. 40â64. Shladover, S. E., D. Su, and X.-Y. Lu. 2012. Impacts of Cooperative Adaptive Cruise Control on Freeway Traffic Flow. Transportation Research Record: Journal of the Transportation Research Board, No. 2342, pp. 63â70. Shoup, D. 2007. Cruising for Parking. Access Magazine, Vol. 36, pp. 16â22. Tsugawa, S., T. Saito, and A. Hosaka. 1992. Super Smart Vehicle System: AVCS Related Sys- tems for the Future. In Proceedings of the Intelligent Vehicles â92 Symposium. Institute of Electrical and Electronics Engineers, New York, pp. 132â137. Tsugawa, S., S. Jeschke, and S. E. Shladover. 2016. A Review of Truck Platooning Projects for Energy Savings. IEEE Transactions on Intelligent Vehicles, Vol. 1, No. 1, pp. 68â77. Varaiya, P. 2005. What Weâve Learned About Highway Congestion. Access Magazine, No. 27, pp. 2â9.
388 NATIONAL COMMITMENT TO THE INTERSTATE HIGHWAY SYSTEM Viegas, J., and L. Martinez. 2015. Urban Mobility Upgrade: How Shared Self-Driving Cars Could Change Urban Traffic. Policy Paper. International Transport Forum, OECD. http:// www.oecd-ilibrary.org/transport/urban-mobility-system-upgrade_5jlwvzdk29g5-en. Viegas, J., and L. Martinez. 2017. Transition to Shared Mobility: How Large Cities Can Deliver Inclusive Transport Services. Corporate Partnership Board Report. International Transport Forum, OECD. https://www.itf-oecd.org/transition-shared-mobility. Wayland, M. 2015. Adaptive Cruise Control Goes Mainstream. Detroit News, March 3. http://www.detroitnews.com/story/business/autos/2015/03/03/adaptive-cruise-control- growing/24352141.