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.
1Welcome and Introductory Remarks Peter Sweatman of the University of Michigan Transportation Research Institute, Ann Arbor, Michigan, welcomed the participants to the sym- posium. He acknowledged the symposium sponsorsâ the Office of the Assistant Secretary for Research and Technology, U.S. Department of Transportation (DOT); the Directorate-General for Research and Innova- tion, European Commission; and the Transportation Research Board (TRB) of the National Academies1â and noted that the symposium was the third sponsored by the three organizations to enhance cooperation and coordination between the United States and the Euro- pean Union. Sweatman suggested that the topic of road automation is of great interest to public agencies, the automotive industry, technology companies, and other diverse stakeholders. Sweatman next recognized and thanked the sympo- sium planning committee and acknowledged the special assistance of Planning Committee Co-Chair Maxime Flament of ERTICO-ITS Europe. He observed that the hard work of the committee members provided an excel- lent example of transatlantic cooperation and noted that the planning committee spent a lot of time developing the use case scenarios and the breakout group format to ensure a productive symposium. He also thanked the authors of the white papers, Richard Bishop of Bishop Consulting; Steve Shladover of the University of Cali- 1 On July 1, 2015, the official name of the National Academy of Sciences became the National Academies of Sciences, Engineering, and Medicine. fornia, Berkeley; Oliver Carsten of the University of Leeds; and Risto Kulmala of the Finnish Transporta- tion Agency. Sweatman recognized Katie Turnbull of the Texas A&M Transportation Institute, who would serve as the symposium rapporteur and complete the symposium proceedings, and Barbara Siegel, who would graphically record the symposium sessions. He praised the excellent support given by Monica Starnes of TRB and Frank Smit of the European Commission. Sweatman stressed the importance of road transport automation on the efficient movement of people and goods. He noted the impacts of recent advancements in vehicle and information technologies. Sweatman chal- lenged symposium participants to actively engage in discussions over the 2 days and to share their ideas on opportunities, challenges, and potential research topics associated with road transport automation. Manuela Soares of the Directorate-General for Research and Innovation, European Commission, noted the impor- tance of transatlantic communication and cooperation to address common transport challenges. She indicated that the first two symposiums had fostered increased dia- log and collaboration between the partners and noted the mutual interest in road automation and the variety of activities under way in Europe and the United States. Soares thanked Sweatman and the planning commit- tee for their hard work in organizing the symposium. She remarked that the white papers provided excellent back- ground information and that the use case scenarios set the
2 T O W A R D S R O A D T R A N S P O R T A U T O M A T I O N stage for the breakout group discussions. She stressed the importance of the breakout groups in identifying poten- tial research topics for transatlantic collaboration. Soares noted that the results from the symposium will be of ben- efit in identifying themes for the 2016â2017 work pro- gram of the European Commission Directorate-General for Research and Innovation. Soares recognized and thanked the other symposium sponsors, the U.S. DOT and TRB, and acknowledged the hard work of the European Commission and TRB staff. She noted the importance of the symposium in identi- fying opportunities for ongoing research collaboration and the importance of continuing the partnership to fos- ter transatlantic cooperation and collaboration. Soares thanked the National Academies for hosting the sym- posium and encouraged participants to share their ideas for needed research and opportunities for collaboration. Kevin Womack of the Office of the Assistant Secretary for Research and Technology, U.S. DOT, recognized the hard work of the planning committee and the support from U.S. DOT leadership and expressed appreciation to the authors of the white papers. Womack also thanked Monica Starnes of TRB and Frank Smit of the European Commission for their assistance to the planning com- mittee and completing all the details for the symposium. He noted the effective use of conference calls and e-mail exchanges by the planning committee in developing the symposium format, the use case scenarios, the white papers, and the breakout discussion group process. Womack recognized the leadership of Gregory Win- free, Assistant Secretary for Research and Technology, U.S. DOT, in developing the overall partnership and in organizing the symposium. He noted that Winfreeâs sup- port had been instrumental in advancing the partnership between the U.S. DOT and the European Commission. He also indicated that the topic of automated roadways is of great interest to the assistant secretary and the department. Womack stressed the importance of identifying opportunities for ongoing collaboration and thanked participants for taking time to attend the invitation-only symposium. He noted the interest of diverse public and private stakeholders in the topic and stressed the impor- tance of the breakout groups in identifying transatlantic collaborative research needed to advance the organized, safe, and beneficial deployment of automated roadways. Neil Pedersen, Executive Director, TRB, welcomed sym- posium participants to the recently renovated historic headquarters of the National Academies of Sciences (NAS). He suggested it was appropriate to be meeting in the NAS Building, as the interdisciplinary nature of road automation requires expertise from multiple fields, not just transportation. He noted that TRB is one of six divisions of the National Research Council, which is the operating arm of NAS, the National Academy of Engi- neering, and the Institute of Medicine.2 Pedersen noted that TRB was pleased to provide sup- port for the three joint EU-U.S. symposia as part of the memorandum of understanding. He indicated that the symposia have been successful in facilitating information sharing and promoting collaborations and suggested that this symposium should generate more opportunities for research collaboration. Pedersen highlighted TRBâs interest in roadway auto- mation and recent activities, including sessions at the TRB annual meeting, conferences, and research projects. He reported that automated vehicles (AV) and connected vehicles (CV) were the first hot topic identified by the TRB Executive Committee as part of its new strategic plan. Pedersen noted that TRB is becoming more strategic in identifying emerging and cross-cutting issues. He indi- cated that the results from the symposium will be of use in defining the TRB agenda on AV-CV as well as in iden- tifying opportunities for EU-U.S. research collaboration. Pedersen observed that TRB has had significant engagement in AV-CV research. He suggested that the widespread interest in the topic was evident in the 25 sessions at the 2015 TRB annual meeting that focused on AV-CV research, demonstrations and pilots, policy implications, and security concerns. One of the sessions featured a discussion of AV-CV activities and opportuni- ties by the chief executive officers of several state depart- ments of transportation. He also noted that at least 70 of the 220 TRB standing committees and task forces have indicated an interest in the topic. Pedersen highlighted other upcoming activities, including the Automated Vehicles Symposium, spon- sored by TRB and the Association for Unmanned Vehi- cle Systems International, on July 20 to 24, 2015, in Ann Arbor, Michigan. The University of Michigan Transpor- tation Research Institute is a cosponsor of this confer- ence. The December 2015 TRB university transportation center conference is also focusing on AV-CV research and deployment. Pedersen highlighted examples of research studies under way through the Cooperative Research Programs. Current projects are examining the legal environment for driverless vehicles, the costs and benefits of public-sector deployment of vehicle-to-infrastructure (V2I) technolo- gies, and the potential impacts of AVs on state and local transportation agencies. Other projects focus on the impacts of transit system regulations on AV-CV intro- duction and AV-CV applications in freight operations. 2 On July 1, 2015, the Institute of Medicine became the National Academy of Medicine and joined the National Academy of Sciences and the National Academy of Engineering as the third academy overseeing the program units of the National Academies of Sciences, Engineering, and Medicine.
3w e l c o m e a n d i n t r o d u c t o r y r e m a r k s Pedersen acknowledged the work of the symposium planning committee. He noted that the use case scenar- ios introduced potential practical applications into the breakout group discussions. He suggested that thanks to the efforts of the planning committee and the spon- sors, the symposium really focused on where hype meets reality and where vision and dreams meet practicality and implementation. Pedersen further suggested the need to focus discussions on the current state of the practice, which is quickly evolving, future directions and possibili- ties, and what can practically be implemented in the near term and the longer term. In closing, Pedersen thanked the symposium sponsors: the European Commission and the U.S. DOT Office of the Assistant Secretary for Research and Technology. He encouraged active participation over the next 2 days, especially in identifying potential research topics for future EU-U.S. collaboration.
4Opening Plenary Session Chris Urmson, Google, Mountain View, California, USA Richard Bishop, Bishop Consulting, Granite, Maryland, USA Steven E. Shladover, University of California, Berkeley, California, USA Oliver Carsten, University of Leeds, Leeds, United Kingdom Risto Kulmala, Finnish Transport Agency, Helsinki, Finland Maxime Flament, ERTICO-ITS Europe, Brussels, Belgium Keynote Presentation: realizing self-Driving Cars Chris Urmson Chris Urmson discussed the work under way at Google related to self-driving cars. He described the interest and motivation at Google for self-driving vehicles, his back- ground and interest in the area, and recent research and tests being conducted by Google. Urmson suggested that the invention of the automo- bile by Carl Benz in 1885 was an amazing step for- ward for society, with a major impact on shaping cities, enabling interstate commerce, and providing mobility. He noted that the first public demonstration of the vehicle ended with Carl Benz crashing it into a wall. He commented that work has been under way to reduce vehicle crashes ever since. Urmson noted that the first crash was symptomatic of a much bigger problem today, with approximately 33,000 roadway fatalities annually in the United States and 1.2 million fatalities worldwide. To put these numbers into context, Urmson indicated that the 33,000 annual fatalities would equal a 737 airplane crashing 5 days a week (every workday). He noted that approximately 94% of these crashes are due to human error, which technology could help address. Urmson noted that traffic congestion is an issue in all urban areas. He suggested that the road system has not kept pace with increases in vehicle miles traveled: between 1990 and 2010, vehicle miles traveled grew by 38%, while the road system grew by only 6%. He sug- gested that only about 8% of the freeway surface area is being used at maximum throughput. He noted that automation would allow for tighter vehicle spacing that potentially would double the maximum throughput. Urmson described the human impact of traffic con- gestion. The average commute in the United States is 50 minutes per worker per day. Urmson noted that when this figure is multiplied by 120 million workers, approxi- mately 6 billion minutes per day are wasted being stuck in traffic. He suggested that reducing the time people spend in traffic would increase productivity and reduce stress. He also noted that alternatives are needed for people who are unable to drive because of physical or financial limita- tions and observed that the aging of the Baby Boom gen- eration will increase the number of individuals who need alternatives to driving. Urmson described a situation of a vision-impaired individual who has a 3-hour commute rather than a 30-minute commute because he is unable to drive. Urmson noted that his teamâs mission at Google is to improve peopleâs lives by transforming mobility. Urmson described his background working in vehi- cle automation, which began with participation in the Defense Advanced Research Project Agency (DARPA) Grand Challenge while he was in graduate school. The Grand Challenge was initiated in response to a congressio- nal mandate that by 2015, one-third of all military ground vehicles would be unmanned. The goal was to advance the rate of technical progress with unmanned vehicles. Urmson explained that the DARPA Grand Challenge was a race of vehicles operated completely autono- mously across the desert from Los Angeles to Las Vegas, a distance of approximately 130 miles. He reported that
5O P E N I N G P L E N A R Y S E S S I O N the first year, Googleâs vehicle traveled 7 miles at peak speeds of 40 miles per hour. He noted that the second year, their vehicle and others completed the race. He described the 2007 DARPA Urban Challenge in which unmanned vehicles recognized and responded to four- way stops at intersections and self-parked. No additional challenges were held, and Urmson noted that the com- munity of people participating in the challenges dis- banded and moved on to other activities. Urmson described Googleâs interest and work in the area. He noted that the self-driving car team was formed in 2009 with a focus on fielding the technology and hav- ing an impact on the world. The two initial goals were to operate self-driving vehicles 100,000 miles on public roads and 1,000 miles on roads with a lot of variation. One of these roads was the El Camino Real from San Jose to San Francisco, California, which has approxi- mately 240 signalized intersections and changes from four lanes in each direction to one lane in each direction. Urmson indicated that several technologies were developed during the initial 18-month project, includ- ing high-resolution maps. Core elements of these maps included a spatial point cloud derived from data captured by infrared lidar, elevation models that provided a three- dimensional shape of the world, and vector representa- tions of where to expect lane markings, traffic signals, crosswalks, and other roadway elements. Urmson noted that these technologies are used to determine the loca- tion of the self-driving vehicles. He provided an example that illustrated the improved accuracy of the system as compared with GPS. Urmson noted that the two goals were accomplished within an 18-month period, after which a decision was made to focus on the next phase of freeway driving. He indicated that the fleet of Prius self-driving vehicles was upgraded to a fleet of Lexus self-driving vehicles. After undergoing additional testing on freeways, these vehicles were made available for employees to use. (Urmson indi- cated that Googleâs practice is to involve employees in product testing.) Urmson noted that the response was very positive, with more than 100 employees using the self-driving vehicles on a daily basis to commute to work and to make other trips. He indicated that the employees loved the technology, even those who were skeptical at the beginning. He also noted that the employees, espe- cially those with long commutes, reported feeling more energized after arriving home in the evening (as com- pared with their normal commute). Urmson noted that part of the assessment focused on what was happening inside the vehicle, explaining that the Google team had anticipated that people might overtrust the technology. Examples of observed behavior included a driver leaning back so that other travelers would think that there was no one in the vehicle and a driver reach- ing into the back seat for a laptop to charge his phone. In noting that these were not desired behaviors, Urmson said that the Google team had three options at this point: launch the technology, spend time debugging drivers, or reevaluate the project goals in the context of the mission to âimprove peopleâs lives by transforming mobility.â Urmson said that after assessing the situation, the team concluded that people want technology that gets out of the way. He suggested that in-vehicle technology that lets people know when they are doing something incorrect is not getting out of the way. He also noted that such technology does not help provide mobility to individuals with special needs. Therefore, the team took a step back and refocused on vehicles that can drive everywhere. He illustrated the point with a photograph of an intersec- tion in Mountain View that includes a traffic signal, an at-grade railroad crossing, and vehicles, bicyclists, and pedestrians moving in all directions. He explained that interpreting the motion of vehicles in this type of setting is much more difficult than driving on a freeway. Urmson reported that the Google team also examined the in-vehicle experience for drivers and the vehicle hard- ware. He suggested that when there is an individual in a vehicle who is being counted on to take over the driv- ing function, there is inherent redundancy in the human operator. He noted that when the vehicle is driving with- out an operator, the human redundancy is removed. As a result, redundancy has to be built into the electronics architecture, the actuation architecture, and the sens- ing architecture. He suggested that vehicles are designed around the driver and that therefore there is a lot more opportunity to be innovative when the car is driving. Urmson discussed the prototype concept vehicle being developed by Google. He noted that the vehicle is limited to speeds up to 25 miles per hour, primarily for safety rea- sons, and that it is being designed to operate in all types of urban settings, with all the challenges involved in the urban environment. Urmson described the pyramid of protection Google is using in designing the vehicle. At the apex of the pyramid is the physical protection of the occupants. Elements of the vehicle include the frame, the wheels, the sensing components that allow it to perceive the world, the electronics and drive train, the electronic interfaces, the interior body, and the exterior crash surface. Describing the power architecture of the vehicle, Urmson noted that while the power comes from the tractive motor, there are redundant power buses and batteries that allow the intelligent components to con- tinue to function to drive the vehicle into a safe state if needed. The operation of the redundant power system and its ability to bring the vehicle to a safe condition were highlighted in a video. Urmson commented that the vehicle has 360-degree laser and radar coverage as well as camera coverage. He described a Google-developed laser that provides 200 meters of vision with a narrow, steerable field of view. He noted the system can detect
6 T O W A R D S R O A D T R A N S P O R T A U T O M A T I O N both a cinder block 150 meters ahead of the vehicle and a bicyclistâs arm gestures and react accordingly. Urmson noted that one criticism of Googleâs vehicles has been the Velodyne laser system mounted on the roof, which costs approximately $75,000. He said that the Velodyne laser has been replaced with a laser developed in-house at a lower cost. Sensors are imbedded around the vehicle. He reported that in combination, these sen- sors provide short-, mid-, and long-range coverage and an unprecedented degree of perception capabilities. Urmson illustrated the prototype vehicle self-driving around the Google test facility in Californiaâs Central Valley. He noted that the vehicle interacts with other vehicles, pedestrians, and bicyclists under daily traffic conditions. He described the process of establishing an initial position of the vehicle by using the map-matching algorithm to align the vehicle very precisely against the map. The vector representation of where lane markings, crosswalks, and other features are is layered on the vehi- cle. The real-time view from the vehicle is also layered on the maps. Urmson provided an example of the prototype vehicle detecting traffic cones blocking a lane and a green light at a traffic signal. He suggested that a system based on the current location of vehicles would not be very useful and said that Google is using a predictive model of where all vehicles are moving. The model tracks vehicles at 10 times per second or more. The trajectory the vehi- cle should follow is calculated and the other vehicles that will influence its speed are identified and tracked. The angle of the steering wheel, the speed, and the braking are calculated and set. He described a video highlighting the prototype vehicle traveling through an intersection. Urmson noted that the prototype vehicle also has to be able to interact with vulnerable road users (VRUs), including pedestrians and bicyclists. He described video illustrating the prototype vehicle operating with bicycles in different situations and noted that with almost a million miles of testing, the vehicles have interacted with pedestri- ans, bicyclists, and other vehicles in numerous situations. That information is used to train classifiers to understand what the vehicles encounter and to build behavior predic- tions. Urmson presented some examples of the unique situations that the prototype vehicles have encountered, including a woman in an electric wheelchair chasing a duck in figure eights in the middle of the road. He noted that the prototype vehicle identifies situations that are anoma- lies and treats them with extra attention by slowing down. Other examples included both vehicles and bicyclists run- ning red lights, vehicles changing lanes abruptly without signaling, and vehicles pulling into traffic unexpectedly. Urmson indicated that the prototype vehicles also must be able to recognize police, fire, and emergency medical services vehicles as well as school buses and public transit buses. He noted that these vehicles have special operating characteristics and requirements for other vehicles. Urmson described Googleâs vision for the technology as helping people get from Point A to Point B and carry out their daily activities without driving. He showed a video of people riding in prototype vehicles and noted the reaction has been very positive, including among individuals who may not be considered early technology adopters. He sug- gested that while there is a broad question about societal acceptance of autonomous vehicles, widespread support might be an easier step than most people imagine. In closing, Urmson raised some provocative ideas to help stimulate discussion at the symposium. The first idea focused on vehicle-to-vehicle (V2V) technology. He suggested that V2V communication is an incredible technology that offers amazing opportunities to share information between vehicles but that it should not be a required predecessor for fully self-driving vehicles. Rather, he said, V2V is a great technology to layer on top of the perception capabilities embedded in vehicles. He further suggested that potentially there can be dif- ferent paths to realizing the same societal benefits. He commented that both V2V and automated vehicle (AV) technology paths have challenges. He suggested that working on all paths may provide the best approach. Urmsonâs second idea focused on digital maps. He pointed out the benefits of maps, especially in assisting with a vehicle verification and validation process. He noted the large number of variables associated with the operation of self-driving vehicles. Maps, he said, provide extra informa- tion beyond what the vehicle can sense. He noted that maps also limit the operating environment, which constrains the verification and validation process to some extent. Third, Urmson suggested that self-driving vehicles would not be realized in incremental stages or by mov- ing up through the driver assistance system. He sug- gested that there is a chasm between driver assistance and self-driving vehicles, noting that part of the chasm related to responsibility. He commented that with a driver assistance system, there is a driver who is respon- sible for the operation of the vehicle. The vehicle may be assuming some of the lower-level controls, but the driver is monitoring the operation at all times. With a fully self-driven vehicle that would provide mobility to visually impaired individuals, the responsibility for safe operation rests with the vehicle. Urmson suggested the technology and the market pressure were very different for the two approaches. He suggested that market pres- sure would force driver assistance systems to be lower in cost and have better precision but not necessarily full recall, whereas self-driving vehicles would need to apply much higher precision and recall. Urmson concluded by encouraging all groups to work together to make self-driving vehicles a reality. He noted that he would like to see self-driving vehicles available for his oldest son, especially given the high crash and fatality rates among teen drivers.
7O P E N I N G P L E N A R Y S E S S I O N Presentation of White PaPer 1: roaD transPort automation as a PubliCâPrivate enterPrise Richard Bishop and Steven E. Shladover Steven Shladover and Richard Bishop summarized White Paper 1, âRoad Transport Automation as a PublicâPrivate Enterprise,â the full text of which is included in this volume as Appendix A. They discussed the diversity of automation concepts, the state of the art and state of the market, and technological maturity. They also described nontechnical issues, business models, and potential roles within the pub- lic and the private sectors and identified research topics for exploration through EU-U.S. collaboration. Shladover noted that it is important to remember the diversity of automation concepts and suggested that this diversity may be an impediment to a mutual understand- ing of the concepts being discussed. He indicated that the white paper uses three dimensions to help in framing an understanding of specific applications: the goals to be served by the automation system, the roles of the driver and the automation system in the driving tasks, and the complexity of the operating environment. Shladover indicated that road vehicle automation sys- tems are not ends in and of themselves but are rather a means of satisfying individual and societal needs and goals. He reviewed possible direct user goals of auto- mation systems, including improving driving comfort and convenience, freeing up time consumed by driving, and reducing vehicle user costs and travel times. Other possible system-level goals are improving traffic safety, reducing travel times, enhancing and broadening mobil- ity options, and reducing traffic congestion in general. Additional potential goals are reducing energy use and pollutant emissions, making more efficient use of existing road infrastructure, and reducing the cost of future infra- structure and equipment. He noted that system designs will be different, depending on the projectâs goals. Shladover highlighted the levels of automation defini- tions developed by the Society of Automotive Engineers (SAE) for SAE J3016. As presented in Table 1, the levels TABLE 1 SAE J3016 Definitions of Levels of Automation Level Description Definition Execution of Steering, Acceleration, and Deceleration Monitoring of Driving Environment Fallback Performance of Dynamic Driving Task System Capability (driving modes) Human Driver Monitors the Driving Environment 0 No automation Full-time performance by the human driver of all aspects of the dynamic driving task, even when enhanced by warning or intervention systems Human driver Human driver Human driver Not applicable 1 Driver assistance Driving modeâspecific execution by a driver assis- tance system of either steering or accelerationâ deceleration that uses information about the driv- ing environment and with the expectation that the human driver performs all remaining aspects of the dynamic driving task Human driver and system Human driver Human driver Some driv- ing modes 2 Partial automation Driving modeâspecific execution by one or more driver assistance systems of both steering and accelerationâdeceleration that that uses informa- tion about the driving environment and with the expectation that the human driver performs all remaining aspects of the dynamic driving task System Human driver Human driver Some driv- ing modes Automated Driving System Monitors the Driving Environment 3 Conditional automation Driving modeâspecific performance by an auto- mated driving system of all aspects of the dynamic driving task with the expectation that the human driver will respond appropriately to a request to intervene System System Human driver Some driv- ing modes 4 High automation Driving modeâspecific performance by an automated driving system of all aspects of the dynamic driving task even if a human driver does not respond appropriately to a request to intervene System System System Some driv- ing modes 5 Full automation Full-time performance by an automated driving sys- tem of all aspects of the dynamic driving task under all roadway and environmental conditions that can be managed by a human driver System System System All driving modes
8 T O W A R D S R O A D T R A N S P O R T A U T O M A T I O N range from 0 (no automation) to 5 (full automation). He noted that the system takes on more of the driving responsibility at the higher levels of automation. He also discussed Table 2, which provides examples of systems and driver roles associated with different levels, and indi- cated that the Level 1 and 2 systems are currently com- mercially available. Shladover noted that the automated driving system, not the driver, monitors the driving envi- ronment at Levels 3, 4, and 5. Shladover discussed the complexity of the operating environment. He indicated that the degree of segregation from other road users is critical in discriminating between the different systems. He noted it was important to remem- ber that totally automated systems without drivers have been operating on exclusive guideways for decades. These systems are in operation at airports and in other areas. Shladover noted that exclusive guideways, dedicated highway lanes, general limited-access highways, pro- tected campuses, special-purpose pathways, pedestrian zones, and urban streets all have different degrees of separation. He indicated that the complexity of traf- fic, including the speed, density, and mix of users, also influences the complexity of the operating environment. Weather and lighting conditions, the availability of infrastructure-to-vehicle (I2V) and V2V data, and the standardization of signage and pavement markings rep- resent additional influencing factors. Bishop described the state of the art in the development of automated driving systems. He noted that examples of highway operation include prototype vehicles driving in lane, changing lanes, and merging. He also noted that there are examples of prototype vehicles driving on a wide range of city streets and navigating signalized inter- sections, roundabouts, and other elements. He indicated that key technology elements of these prototype vehicles include sensors (radar, stereo and mono cameras, lidar), data-processing systems, and dynamic maps. Bishop suggested that at Level 4, automated chauf- feuring is viewed as a natural evolution by some original equipment manufacturers (OEMs) and is being pursued by Google, Uber, and others. He noted that these efforts focus on street-level automated driving at low speeds and in limited geographic areas. He also indicated that proto- types of Level 3 automated truck platooning have been demonstrated by some OEMs with a focus on long-haul freight transport on well-structured highways. Bishop discussed the state-of-the-market section in the white paper. He noted that active safety systems, which form the technological foundation for AVs, are currently available on many vehicle models in Europe and North America. He said that examples of Level 2 highway use systems currently available on a few models include simultaneous adaptive cruise control (ACC) and lane centering at highway speeds on well-structured highways with limited curvature. He described Traffic Jam Assist, which provides low-speed automated lateral and longitu- dinal control. With the Traffic Jam Assist system, drivers are instructed to keep their hands on the steering wheel, or the system disables the feature. With combined ACC and lane centering, some manufacturers require hands on the wheel while others do not. He stressed that this will be an important factor to monitor as products evolve, one that could have safety implications. Bishop noted that Level 2 and 3 highway use systems are anticipated to be available by the end of the decade. He noted that these systems will have full speed range and will be capable of accommodating a full range of normal highway curvatures. He further indicated that some approaches will actively monitor the driverâs attention and gaze and provide a warning if the driver does not have his or her eyes on the road. According to Bishop, some systems will simply drive the vehicle in a particular lane, while others will also make lane changes as needed. He noted that OEM announcements and dates for these systems include Toyota by middecade, Audi and GM by 2016, Nissan by 2018, and BMW by 2020. He also described aftermarket systems, including those being developed by small start-up companies. Bishop described the Volvo Drive Me project, which is a Level 3 highway use field test involving 100 vehi- TABLE 2 Examples of Systems at Each Automation Level Level Example System Driver Role 1 Adaptive cruise control or lane-keeping assistance Must drive âotherâ function and monitor driving environment. 2 Adaptive cruise control and lane-keeping assistance Traffic Jam Assist (Mercedes) Must monitor driving environment. (System nags driver to try to ensure monitoring.) 3 Traffic Jam Pilot Automated parking May read a book, text, or web surf but be prepared to inter- vene when needed. 4 Highway driving pilot Closed campus driverless shuttle Driverless valet parking in garage May sleep; system can revert to minimum risk condition if needed. 5 Automated taxi (even for children) Car share repositioning system No driving needed.
9O P E N I N G P L E N A R Y S E S S I O N cles for use by the public on limited to specific roads in Gothenburg, Sweden. The system is anticipated to be operational by 2017. Bishop indicated that automated valet parking, despite being a Level 4 system, will come to market quickly because vehicles are within a parking lot or parking garage and are traveling at low speeds. He noted that Level 4 automated chauffeuring is being tested in Europe and that Google has indicated that pilot testing of the system will begin this year. Bishop observed that truck platooning has received a lot of attention recently. He noted that truck platoon- ing, which involves longitudinal control only, is a Level 1 system. The combination of radar and V2V enables vehicles to follow at less than 100 feet. He suggested that the vehicle drafting and aerodynamics provided by truck platooning results in substantial fuel economy ben- efits that make it compelling to the trucking industry. He noted that commercial offerings are expected within the next 2 to 3 years, with pilot testing in the U.S. likely to begin this year. In summarizing the state of the market, Bishop sug- gested there were two parallel paths: everything some- where and something everywhere. He credited Bryant Walker Smith for these terms. Bishop described the everything somewhere path as full automation in some applications in some locations. The Google car and the CityMobil projects are examples of this path, he sug- gested. Bishop noted that this path may focus on fleet operations, which involve frequent servicing and test- ing to ensure safe operation. He suggested that every- thing somewhere is a viable path, although it may be limited geographically. The second path, something everywhere, involves a more limited level of functional automation that can be used everywhere. He suggested that this path follows the classic incremental approach whereby systems are brought to market on private vehicles capable of operating on any road, with no geo- graphical limitations. Bishop noted that the importance of infrastructure support for automation product introduction is under debate. He said that while some feel that product intro- ductions will proceed without requiring infrastructure support, clearly some level of infrastructure support will be essential to gain transportation benefits. He described the various types of support that may be needed, includ- ing I2V and V2V real-time data, physical protection from hazards, and digital infrastructure for static and dynamic data. Other types of support include sensor-friendly sign- age and markings, better lighting, and higher mainte- nance standards. Bishop highlighted the scenarios for providing sup- port discussed in the white paper. One scenario focuses on private providers with little support from public agen- cies. In a second scenario, the automobile industry and users push public agencies to prioritize this support for different applications. In a third scenario, public agen- cies provide support proactively on the basis of perceived public benefits. Shladover discussed the diverse stakeholders involved in developing, testing, and deploying AV-CV. These stakeholders include vehicle manufacturers and suppli- ers, other technology industry companies, regulators and public authorities, infrastructure and road opera- tors, and public transport operators. Other stakehold- ers are the goods movement industry, users and private drivers, VRUs, and shared vehicle and fleet opera- tors. A final set of stakeholders identified in the white paper includes the insurance industry, big data service providers, research and academic institutions, and the legal profession. Shladover noted that the white paper describes the different types of issues relevant to the various stakeholders. Shladover indicated that the white paper exam- ines the advances in technology that will be needed to achieve Level 3 and above applications, and research opportunities to address these challenges. The white paper presents these technology challenges by degree of difficulty. He noted that the first category included technologies that need some development but no funda- mental breakthroughs. He indicated that technologies included in this category were wireless communications and localization. He noted that the next level of more challenging categories address a series of human factors and driver interface challenges, including safe control transitions, deterring misuse and abuse, encouraging vigilance, and facilitating correct mental models of sys- tem behavior. The white paper also addresses cyber- security, but Shladover noted that cybersecurity is an issue for any modern vehicle with on-board electronics and is not unique to AV-CV. Shladover described some of the even more challeng- ing issues, including fault detection, identification, and accommodation within cost constraints. He suggested that ethical considerations in computer control repre- sented another challenging issue. Environment percep- tion and threat assessment, including minimizing false positives and false negatives under diverse conditions with affordable sensors, represent another challeng- ing issue identified in the white paper. Software safety, including the designing, developing, verifying, and vali- dating of complex software systems to a higher level of safety, was the final and most challenging technology issue discussed by Shladover. He suggested that research is needed on the appropriate mix of formal methods, simulation, and testing to address these topics. Shladover highlighted some of the nontechnological issues associated with AV-CV deployment identified in the white paper. Examples included public policy, legal issues, vehicle certification and licensing, public accep- tance, insurance, and assessment of benefits and impacts.
10 T O W A R D S R O A D T R A N S P O R T A U T O M A T I O N He noted that these issues may be more challenging to address on a transatlantic basis because of the differences in public institutions and public policies. He suggested that there is a need to work harder to identify common lessons from addressing nontechnical issues in the Euro- pean Union and the United States. Shladover suggested that business models and publicâ private roles may also be different between the European Union and the United States because of their different insti- tutional structures. He described the standard approach of privately owned vehicles operating on public infra- structure with limited interaction. Shladover indicated that automation may have synergistic benefits from closer coupling of vehicles and infrastructure, which might open up integrated business models with common ownership of vehicles and infrastructure providing transportation as a service. He suggested that this model mirrors the railroad industry. He reviewed the financing options examined in the white paper, including joint publicâprivate financing, road user charges, new publicâprivate partnerships, and investments from the information technology industry. Shladover indicated that the white paper identifies a wide range of research needs to address technology issues and nontechnical concerns. He highlighted a few of the nontechnical research issues, including possible changes in driver licensing and testing requirements, potential regulations of AVs at the national and state levels, and the need for more uniform standards for roadways and roadside infrastructure. In closing, Shladover described some of the major unresolved questions. He noted that some of the ques- tions are philosophical, while research could help address other questions, including the following: ⢠How much support and cooperation do AVs need from roadway infrastructure and other vehicles? ⢠How big a public sector role should be provided in infrastructure support? ⢠Do higher levels of automation require fundamen- tal breakthroughs in some technological fields? ⢠What roles should national, regional, and state governments play in determining whether an AV is safe enough for public use? ⢠How safe is safe enough? ⢠How can an AV be reliably determined to meet any specific target safety level? ⢠Should AVs be required to inhibit abuse and mis- use by drivers? ⢠Are new publicâprivate business models needed for higher levels of automation? ⢠How will AVs change public transport services, and will societal goals for mobility be enhanced or degraded? ⢠What will be the net impacts on vehicle miles trav- eled, energy, and the environment? Presentation of White PaPer 2: roaD transPort automation as a soCietal Change agent Oliver Carsten and Risto Kulmala Oliver Carsten and Risto Kulmala summarized White Paper 2, âRoad Transport Automation as a Societal Change Agent,â the full text of which is included in this volume as Appendix B. They described the potential benefits of road transport automation to individuals and to society. They also discussed the potential short- and long-term costs to individuals, vehicle owners, infra- structure owners and operators, service providers, the automotive industry, and authorities. Carsten noted that the initial focus of the white paper was on the potential socioeconomic impacts of automa- tion and the groups receiving the benefits and bearing the costs. He indicated that the white paper was expanded to include broader societal implications of automation. Carsten presented photographs from the 1950s and from 2015 illustrating the same concept of self-driving automobiles. He noted that in both cases, the vehicles were envisioned to travel at high speeds on roadways without any driver interaction, allowing drivers and pas- sengers to engage in infotainment. Carsten cited vehicles from Daimler and Google as current examples of this approach, along with the CityMobil2 urban transit vehi- cles and the truck platooning operations. He noted that the white paper examined two time frames: the incre- mental near and medium term, or the next 5 to 10 years, and the long term or transformational period. Carsten reviewed the potential individual benefits of road transport automation. He suggested that access to infotainment appears to represent one of the major ben- efits. He noted that the potential to work in a vehicle rather than drive has a major impact on both the value and the cost of travel time. He indicated that in the United Kingdom, an individual spends approximately 235 hours a year driving. He noted that road transport automa- tion has the potential to result in major lifestyle changes that will improve the quality of life. He also noted that long-distance commuting by private vehicles may become more palatable to individuals and thereby make possible a wider choice of residency location. Other potential ben- efits highlighted in the white paper included the reduced risk of fines related to compliance with traffic laws and regulations, increased comfort in driving, potential cost savings from increased safety and reduced insurance pre- miums, and accessibility of driving for elderly individuals. Carsten suggested that technical equipment on Level 3 and 4 vehicles would provide substantial safety ben- efits even when used in manual driving and at Levels 1 and 2. The white paper examines estimates of reduc- tions in crashes and fatalities resulting from automated
11O P E N I N G P L E N A R Y S E S S I O N road transport. One source, eIMPACT, estimated that the benefits could be on the order of a 50% reduction in fatalities. Roadway efficiency and capacity should also improve, especially with cooperative intelligent transpor- tation systems, which should increase vehicle through- put. Carsten also suggested that there could be negative consequences for nonautomated vehicles and other road users. For example, long truck platoons could form moving roadblocks for other users. In addition, there is the potential in urban areas for pedestrians and cyclists to lose road space. He noted that this issue came up in the CityMobil2 test in La Rochelle, France, as bicyclists lost road space for the track of the automated bus. He suggested that with regulation, cooperative intelligent transportation systems could help to address problems of interaction between AVs and manual vehicles. Carsten reviewed some of the environmental ben- efits of road transport automation outlined in the white paper. He noted that vehicles operating under automated control should be more fuel-efficient, which will reduce energy consumption and emissions. He also noted that AVs could encourage more long-distance driving, more long-distance commuting, and more urban sprawl. Kulmala discussed the potential costs to different groups from road transport automation. He suggested the costs to individuals may include special driver training and road user education and special licenses or permits to operate an AV. He suggested that the costs of owning and maintaining an AV would be higher than those for conventional manually driven vehicles and reported that todayâs technology packages for Level 2 automation were approximately $3,000. He noted that the higher degree of redundancy for safety-critical systems and components in AVs increases their costs. He also discussed the willing- ness of vehicle owners to pay for driver support systems and for self-driving capabilities. Kulmala described some of the potential costs for infrastructure owners and operators. He suggested that special lanes or roads reserved for AVs might be needed if a critical mass of AVs existed. He noted that repurpos- ing or redesignating existing dedicated lanes could also occur and that road markings and traffic signs would need to be harmonized, visible, and in good condition. Kulmala used Figure 1 to illustrate the variety of signs for pedestrian crossings in use today. He also noted that winter maintenance is a concern in many countries and regions. Other possible costs for infrastructure owners and operators highlighted by Kulmala included roadside features such as landmarks, posts, and poles to facilitate automated driving in adverse weather and on private roads. He indicated that the availability of infrastructure for I2V and vehicle-to-infrastructure (V2I) communica- tions, including dedicated short-range communication and cellular communication, may influence costs for operators. He further suggested that establishing a sys- tem for recovering costs through taxes, road user charges, or usage fees may be needed. Kulmala suggested that one of the key costs for ser- vice providers was digital maps with sufficient quality for self-localization and environment interpretation. He noted that local dynamic maps are needed to collect information for vehicle decision making and indicated that additional sensors may also be needed to provide an electronic horizon. He described the need for data on basic road features as well as road malformations such as potholes and ruts. Kulmala discussed the need for FIGURE 1 Examples of âPedestrian Crossingâ signs. [SourCe: VRUITS (Improving the Safety and Mobility of Vulnerable Road Users Through ITS Applications), 2014.]
12 T O W A R D S R O A D T R A N S P O R T A U T O M A T I O N high-quality, real-time traffic information, especially for events, incidents, and congestion. In describing the potential impacts on the automotive industry, Carsten noted that the costs associated with vehicle manufacturing may increase as a result of the complexity of the basic elements of automated driving. Examples of these elements include extended environ- mental sensing, accurate positioning, vehicle-to-every- thing (V2X) connectivity, the need to preserve driver and occupant privacy, and the need to ensure security. He suggested that these costs may decrease over time with the mass production of AVs. He further suggested that there may be additional costs related to standardization, training vehicle dealers, and vehicle servicing. Carsten noted that AVs are complex and may utilize proprietary technology. Finally, he suggested that insurance-related costs are likely to be affected if liability for vehicle opera- tion at higher levels of automation is transferred from the driver to the vehicle manufacturer. Kulmala described possible costs to authorities, includ- ing the need to establish regulations concerning AVs. He suggested that this process would require resources to conduct needed research, examine cross-border har- monization, and develop needed regulations. He noted that there may be a shift in the liability of stakehold- ers in the case of crashes. He commented that topics to be examined in this area included product liability and liability defenses; contributory negligence; misuse of a vehicle; self-certification processes by the vehicle indus- try, including tests; and use of event data recorders. Other topics included data and privacy protection and ownership of and the right to use the data produced by AVs. Kulmala noted that theft and security measures will also be required to prevent vehicle theft and hacking, just as with non-AVs. Other items that might increase costs for authorities included certification and roadwor- thiness testing, the need for standardization of vehicle performance (acceleration, braking, time headway, and response lag), as well as methods of informing the driver to take control back from the vehicle. He suggested that there may be a need for a global agreement on infrastruc- ture requirements. Carsten discussed some of the long-term benefits associated with AVs. He noted the transformational potential of automated driving as a new mode of trans- port and suggested that AVs could be as revolutionary as the introduction of the automobile at the turn of the 20th century. He suggested that AVs have the potential to reduce individual vehicle use and increase rideshar- ing. Carsten noted that issues associated with personal security will need to be addressed. He commented that if travel in AVs is too convenient, the use of AVs could be partly at the expense of walking and cycling. He also sug- gested that AVs could have a large impact on logistics, including the potential for last-mile delivery, and that AVs may have positive and negative effects on employ- ment, especially for truck, bus, delivery, and taxi driv- ers. Other positive impacts Carsten noted were the need for less space in which to park vehicles and increased access to employment for individuals who currently do not have a vehicle available. He suggested that potential negative impacts include increases in long-distance com- muting and residential dispersion and increases in road freight while other freight modes were used less. The benefits of AVs for individuals highlighted by Carsten included mobility for those who do not have a vehicle or a driving license as well as for those with physical impairments. Other individual benefits included increased efficiency in time gained from vehicles that park themselves and more affordable mobility resulting from lower levels of vehicle ownership and increased subscriptions to vehicle sharing or ridesharing. Addressing the potential social benefits associated with AVs, Carsten cited the use of travel time for work and entertainment, which has implications for the value of travel time used to calculate the benefitâcost analy- sis of transport-related investments. The potential safety benefits would include the replacement of the driver with more reliable systems that would not be subject to alco- hol abuse, fatigue, inattention, or distraction. Carsten further suggested that AVs would comply with traffic regulations and that I2V and V2V would increase the safety of driving in conditions of poor visibility. Another potential social benefit discussed by Carsten was a substantial increase in roadway efficiency and capacity, depending on the extent of continued manual driving, which would require the need for management of the interaction of manual vehicles and AVs. He sug- gested that narrower dedicated lanes for AVs could lead to increases in capacity and commented that consider- ation may need to be given to how motorcycles would be accommodated on these lanes. Carsten also noted that AV applications might reduce public transport costs. He cited reductions in energy consumption (as a result of smoother driving and fewer incidents), reductions in vehicle emissions, and reductions in land needed for parking as possible environmental benefits that would derive from AVs. Kulmala described possible longer-term costs. He noted that public information campaigns and awareness measures may be needed both for individuals utilizing AVs and for drivers of nonautomated vehicles, cyclists, pedestrians, and other travelers. He commented that developing an awareness of the behavior of AVs among other users would be important. Kulmala indicated that the cost of fully automated vehicles is likely to be higher than that of nonautomated vehicles, but that if vehicle sharing or leasing is used, individuals may not need to purchase vehicles. As a result, actual use costs may be lower. He suggested that
13O P E N I N G P L E N A R Y S E S S I O N in the long term, if all vehicles were fully automated, they might be lighter and simpler than todayâs vehicle, which would result in lower costs. Kulmala noted that the potential long-term costs to infrastructure owners and operators would be higher owing to more widespread AV applications. He sug- gested that changes in road paving and repaving prac- tices, including the use of higher-quality and more expensive aggregate and new paving equipment, might be needed because of narrower lanes and stricter lane keeping. He also noted that AVs may require higher asset management standards related to road pavement condi- tions, signs, and markings and suggested that consider- ation may also need to be given to other changes in road infrastructure, observing that, for AVs, roundabouts are more efficient than traffic signals. Restriction of urban zones to automated public transport, pedestrians, and bicyclists may also need to be considered. Kulmala suggested that higher-quality maps and ser- vices may be needed in the long term, which will increase costs to service providers. Further, he noted that there will still be a need for towing and roadside breakdown services. He commented that higher service levels would likely be needed for AVs, but that V2X and accurate positioning data would assist in helping provide road- side services. Kulmala suggested that AVs might lead to major changes in the automotive industry over the long term. He commented that, on the one hand, full road trans- port automation might result in fewer vehicles in use and thus reduce the automotive industryâs profits, but on the other hand, more intensive vehicle use and the servicing of AVs might result in increased profits. He suggested that relationships with service providers may become more important in the future than relationships with individuals, and that the industry focus may change from vehicle manufacturers to the service providers. Kulmala suggested that the long-term costs for authorities were similar to the near-term costs, focus- ing on regulations, liability, and safety and security. He noted that further consideration may need to be given to ensuring that driverless vehicles are not used to commit crimes or as weapons of destruction. In concluding, Carsten noted that common themes in the white paper included the substantial requirements that AVs would place on road operators and the need for regulations governing the design, operation, and use of AVs. He suggested that the benefits to individuals and to society would be substantially increased with coopera- tive intelligent transportation systems and with increased management of the road system. Topics for further dis- cussion, he said, include the potential for more sharing of vehicles in Europe but more private ownership of vehi- cles in the United States and the impacts of road trans- port automation on land use and development patterns. setting the stage for the symPosium Maxime Flament Maxime Flament, vice chair of the symposium planning committee, introduced and thanked the members of the symposium planning committee, whose names are listed in Table 3. He also thanked the authors of the white papers and the staff from the U.S. Department of Trans- portation, the European Commission, and TRB for their assistance in organizing the symposium. Flament noted that the planning committee spent a good deal of time at its first meeting discussing the focus and mission for the symposium. He indicated that the title of the symposium, âTowards Road Transport Automation: Opportunities in PublicâPrivate Collabora- tion,â reflects the results of this discussion. The planning committee felt there was a need to better understand the roles and responsibilities of the private and public sectors in advancing road transport automation and to identify areas for EU-U.S. collaboration. Flament reviewed the mission statement for the sym- posium developed by the planning committee: âWhat are the complementary roles and responsibilities of the actors in a publicâprivate ecosystem needed to drive the evolution of automated vehicles toward a 21st century mobility system (integrating and optimizing vehicle, user, and infrastructure)?â He noted that the words in bold reflect the key elements of the mission statement. Flament reviewed the desired outcomes for the sym- posium: first, to foster transatlantic partnerships and future collaboration on research areas of mutual interest and, second, to draw out research challenges worthy of international collaboration. He noted that these poten- tial research topics could be advanced by the symposium sponsors and other groups, including the U.S. Department of Transportationâs Intelligent Transportation Systems Joint Program Office, the National Cooperative High- way Research Program, and the European Commission. Flament discussed the three axes of the symposium: the constituencies (he observed that the symposium par- ticipants represented a mix of public- and private-sector TABLE 3 Symposium Planning Committee United States European Union Peter Sweatman, University of Michigan Transportation Research Institute, Chair Maxime Flament, ERTICO-ITS Europe, Vice Chair David Agnew, Continental Automotive NA Roberto Arditi, SINA Group Robert Denaro, ITS Consultant Aria Etemad, Volkswagen AG Ginger Goodin, Texas A&M Transportation Institute Natasha Merat, University of Leeds
14 T O W A R D S R O A D T R A N S P O R T A U T O M A T I O N constituencies, as illustrated in Table 4), the key topics (the subject of the white papers), and the use case scenar- ios that would serve as the basis for the breakout groups. He noted that key topics of discussion in each of the use cases focused on technology, legal issues, business mod- els, and security. Other topics included human factors, policy making, testing, and user acceptance. Flament discussed the three use case scenarios, includ- ing the basic components presented in Table 5: Use Case Scenario 1. Freeway Platooning: Moderately Automated Freeway Operation, Use Case Scenario 2. Automated City Center: Highly Automated Urban Operation, and Use Case Scenario 3. Urban Chauffeur: Fully Auto- mated Tailored Mobility Service. He noted that the use cases had been distributed to the participants prior to the meeting and described the dif- ferent characteristics (level of automation, speed, need for dedicated space, public or private lead, and proj- ect examples) associated with each use case. Flament explained that the use case scenarios were developed by the planning committee to highlight potential applica- tions at different levels of automation that serve different markets and use groups and that reflect different time frames for implementation. Flament described the use case scenario breakout group process illustrated in Figure 2 and the logistics for the breakout sessions. Each scenario would be presented in a plenary session by the planning committee champions. The symposium participants would then convene in their assigned breakout groups for 70 minutes. There were six groups, each of which contained between eight and 10 participants representing a mix of EU and U.S. attendees and a balance of constituencies. Each breakout group had a facilitator and a recorder. Within their groups, the par- ticipants would review the use case scenario and discuss the opportunities created for different constituents and the barriers limiting deployment. Opportunities for col- laborative EU-U.S. research to address the barriers and advance deployment would be identified and discussed. After the breakout group discussions, the participants would reconvene in a plenary session to hear the facilita- tors and recorders report on the barriers, opportunities, and EU-U.S. collaborative research topics discussed by each group. Flament noted that the composition of the breakout groups was different for each of the three use case scenarios to provide opportunities for interaction with different symposium participants. TABLE 4 Symposium Constituencies Constituency Number of Participants Automotive companies 8 Public authorities 5 Infrastructure, road operators 6 Public transport 3 Goods transport 3 Users, drivers, vulnerable road users 2 Shared vehicles and fleets 1 Insurers 2 Service providers 4 Research 12 Legal 2 TABLE 5 Characteristics of Use Case Scenarios Use Case Level of Automation (SAE) Speed (mph) Dedicated Space Private or Public Examples Available Now (Projects) Interaction with Infrastructurea 1. Freeway Platooning 2â3 High (>70) Possibly both Both Sartre, Peloton 3 2. Automated City Center 3â4 Low (10â40) No Both AdaptIVe 4 3. Urban Chauffeur 4 Low (<30) Both Public Google, CityMobil2 5 a1 = low, 5 = high.
15O P E N I N G P L E N A R Y S E S S I O N FIGURE 2 Process for use case scenario breakout groups. Opening Reminder of Objectives (5 min) Barriers Ideas for Research 1) Opportunities 2) Barriers 3) Research Ideads (for EUâU.S. Collaboration Discussions (20 min each) Q: What EUâU.S. collaborative research ideas could address the Barriers to the use case? Q: What barriers are in the way of the use case and the potential Opportunities (for different constituencies)? Q: What opportunities are created by the use case for different constituency areas? Breakout 20 min 45 min Plenary Breakout Report Opportunities Use Case Introduction
