Impacts of Laws and Regulations on CV and AV Technology Introduction in Transit Operations (2017)

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.

TECHNOLOGY 4 T E C H N O L O G Y

TECHNOLOGY 5 State of AV Technology Development Advances in research and development of AV technology are being announced almost daily, and industry perception is continually changing for even the most knowledgeable people in the field. New announcements or developments could substantially change these contents as time progresses, particularly regarding the technology availability timeline. There are parallel research and development processes occurring between AV, which hold the promise of driverless operations, and connected vehicle (CV) communication technologies that enable safer and more efficient driving for both human- and computer-driven vehicles through warnings and detailed information sharing. Automated Vehicles – More than 20 years ago, AV technology advanced in the U.S. via the USDOT (United States Department of Transportation) Intelligent Vehicle Highway System (IVHS) Automated Highway System (AHS) program (although many research initiatives preceded this technology demonstration dating back to the 1950s)2. In the early 2000s, development was reinvigorated by the DARPA Grand and Urban Challenges, which brought universities and private sector teams together3. AV technology today is generally advancing under private sector initiatives of the automobile industry original equipment manufacturers (OEMS), Tier 1 suppliers, software companies such as Google, robotics-oriented start-up companies, and combinations thereof. Major recent strides in accuracy, affordability, and capability of sensors, software, computing, and geo-location technology are enabling AVs. A few OEMs are actively developing marketable automated vehicle models for the industry shown in Figure 1, and almost all major automobile manufacturers are racing to bring these new product offers to the market place as soon as possible. 2 http://onlinepubs.trb.org/onlinepubs/sr/sr253.html 3 https://en.wikipedia.org/wiki/DARPA_Grand_Challenge

Figure 1. Automated Roadway Vehicles Will Be on the Market by 2020 TECHNOLO s GY 6

TECHNOLOGY 7 Connected Vehicles – Over the last 20+ years, CV technology was primarily driven by USDOT initiatives. Some of the CV program evolution was in direct response to the numerous challenges of the grand vision of the AHS. CV technologies use wireless communications between vehicles, the infrastructure, and mobile devices to improve safety and mobility and reduce environmental impacts of human-operated vehicles4. NHTSA released an advanced notice of proposed rulemaking in August 2014 requiring dedicated short-range communications (DSRC) 5.9 GHz communications capability as a standard for light vehicle manufacturers and is expected to do the same for heavy-duty trucks and buses in the year following5. As of 2016, these mandates have not been made, but are still considered imminent. CV communications can also use 3G and Fourth-Generation Long-Term Evolution (4G/5G LTE) technology for non-safety-critical applications. Vehicle-to-vehicle (V2V) and vehicle-to-other road users (V2X) technologies can improve safety by warning bus drivers of obstacles and imminent crash threats. Vehicle-to-infrastructure (V2I) applications can improve both transit vehicle travel efficiency and passenger service. CV technologies have been used for over 20 years now in hundreds of locales around the U.S. and the world for providing priority green time at traffic signals, known as transit signal priority (TSP). Effectively implementing connectivity through V2V and V2I communications requires USDOT, state departments of transportation (DOTs), and local agency coordination, communication standards, OEM cooperation, and potentially international governmental coordination within the global automobile, transit, and commercial vehicle markets. Technology readiness was demonstrated in Ann Arbor, MI by University of Michigan transportation research institute (UMTRI) in the USDOT Safety Pilot program6. Large-scale field tests of CV applications in the U.S. are scheduled for 2018 in Tampa, FL; New York, NY; and the State of Wyoming, some of which include transit applications. These efforts are all ongoing and will not be addressed in this document7. Automated Transit Vehicles – The prospects for AV transit applications in general mixed traffic operation now appears realistic in the foreseeable future. Automated steering, throttle/propulsion and braking, and precision docking controls for buses have already been demonstrated to improve safety and efficiencies of buses augmenting the skills of human drivers8. Automated transit systems on fixed guideway facilities have been in use for over 40 years9. These transit systems have sophisticated supervisory control functions (connected technologies) necessary for safe and efficient management of even just a single transit line with a small number of individual vehicles. We foresee the marriage of the CV and AV worlds to enable truly driverless transit vehicles in the long term, with corresponding enabling developments in transit station and fixed facility design. In addition, we believe AV transit 4 http://www.its.dot.gov/landing/cv.htm 5 http://www.nhtsa.gov/Research/Crash+Avoidance/Vehicle-to-Vehicle+Communications+for+Safety 6 http://www.its.dot.gov/safety_pilot/ 7 http://www.its.dot.gov/pilots/ 8 http://www.path.berkeley.edu/sites/default/files/documents/IM_15-1_low%20%282%29.pdf 9 https://en.wikipedia.org/wiki/Automated_guideway_transit

applicatio begin to Figure Defin Much dis business of service this priva referred t basic bus services franchise service. Although demand mile” ser (FTA) as AV Fixed traditiona regardles schedule passenge AV On-D automate combinat 10 https://w ns will prov be included 2. Parallel AV itions o cussion is u model” of a s through a te business o this as a “ iness mode and establis to a private fixed-route corridors as vice will like a “mobility o -Route Tra l customer s s of deman s/headways rs indicate emand Tra d dispatchin ion. Rides a ww.transit.d ide operatio in planning /CV Develop f Busi nderway in utomated ca “public tran model is a t private tran ls. A public h a formal p operator to public trans the basic b ly develop, w n-demand” nsit Servic ervice with d, but witho and vehicle their intent t nsit Service g and man re shared a ot.gov/fundin nal flexibility studies as s ment Paths ness M the transpor r services ( sit business ransportatio sit” service. transit entity ublic/private use public it service wi ackbone of hich is bein type of tran e – Commo a fixed set o ut a human s on some o board or a – Common agement of mong any m g/applying/no resulting in hown in Fig Potentially Co odels tation indus e.g., automa model.” Cu n network c There are s could cont partnershi assets (i.e. l ll certainly c public transi g described sit10. n transit ser f origin/des driver. This routes could light the ve transit serv transit vehic ember of th tices-funding a whole pa ure 2. ming Togeth and Se try about the ted Uber o rrent termin ompany (TN everal hybri ract with a p p(s) where t and/right-of ontinue into t service, a by Federa vice within u tinations vis fixed-route s potentially hicle. ice within u les for a use e public. Al /mobility-dem radigm shift er for AV Tra rvice T place for t r Lyft), and t ology being C), and so d combinatio rivate vehic he public en -way) in pro perpetuity a new type of l Transit Adm rban areas ited by the t ervice wou only make s rban areas r-selected o though solo and-mod-sa TECHNOLO that must n nsit Operatio ypes he “private he differenc used to add me have ns of these le operator f tity provides viding the tr long very h “first-mile/la inistration providing ransit vehic ld operate o tops when that provide rigin/destin rides are ndbox-progra GY 8 ow ns es ress or a ansit igh- st- le n set s ation m

TECHNOLOGY 9 possible if the demand is very low for a given origin/destination pair and for a given time of day, this will not be guaranteed. The service operates much like a horizontal elevator. New passengers may enter or exit at any point on a route. This public transit service would have specified pick-up/drop-off locations designed for passenger convenience, which may number many more locations than existing fixed-route transit service. This service may provide specialized (i.e. minimized delay) routing of a given vehicle through the network of routes given the origins and destinations of the riders. AV Paratransit or Rural On-Demand Transit Service – Working from the concept of paratransit as it is provided today, each registered user will be able to pre-define their personal pick-up and drop-off locations and time of day they will be taking their trip. Trip details are customizable to that specific user’s personal needs, and changes are possible in real time as necessary. Advance reservations will be required only shortly (i.e. an intentionally vague time frame that could range from minutes to hours) before the trip actually occurs (depending on vehicle availability and trip location). When the trip reservation includes service involving a disability that requires special attention with human oversight of the boarding and alighting process (e.g., conditions where special equipment or visual/audio attention is required to assist the passenger), this may be accomplished via remote viewing/control of the vehicle and its special equipment by transit system personnel located in the operations control center or an onboard “customer service agent.” AV “Automated Taxi” On-Demand Service – An on-demand vehicle service provides customized rides which may or may not include ride sharing (as determined by pricing and user preferences). Passenger pick-up and drop-off would be determined by the user – e.g., my home driveway, my airline terminal curbfront by “door #3,” my specific destination street address at the curb, and so on. These vehicle services operate like taxi services operate today.

TECHNOLOGY 10 Historical Context of Automated Transit Systems Automated transit systems on fixed guideways have been in operation for more than 40 years. Automated Guideway Transit/Automated People Mover Systems – The USDOT federally funded pilot project of the first fully automated guideway transit system began passenger service in 1964 at Pittsburgh’s South Park (see Figure 3). Following this prototype system’s demonstration of automation viability came the airport systems at Tampa Airport and Seattle/Tacoma Airports. These initial systems became known as automated guideway transit (AGT) systems, but the more whimsical automated people mover (APM) moniker soon became commonly used. Throughout the 1970s and 1980 the successful deployment of fully automated train systems began to allow fundamentally different configurations of airport terminal facilities to be created, such as the massive Atlanta airport with its spine APM system connecting numerous airside concourses. The Atlanta system has now been expanded to carry secure air passengers between destinations spread over more than a mile across the airfield. Many more airport APM systems have been built by numerous system suppliers in the following years. A USDOT federally funded demonstration of the first urban APM application was the initial “loop” system in downtown Miami that began service in the mid-1980s. This “Metromover” was extended in the 1990s to connect several adjacent business districts with the Central Business District (CBD) and provide access points at over 20 station locations. Other urban systems were soon in service throughout the world, beginning with the fully automated urban system in Lille, France, which began service in 1983 and was followed by systems throughout Europe and Asia. One of the first regional-scale automated systems was the Vancouver Sky Train, which began fully automated passenger service in the mid-1980s and expanded several times to include multiple lines. Many fully automated metro systems are now in service throughout the world, such as the Singapore Metro subway system that runs without an operator or even attendant transit personnel onboard11. In general, the term APM is commonly applied to airport and special-use systems, and the term AGT is commonly applied to larger urban systems that reach a full regional/metropolitan scale of service. The last 50 years of AGT/APM system development has also provided a strong platform from which AV transit applications can extend. Fully automated, driverless trains have been safely operated over many millions of vehicle-miles with no service-related passenger fatalities. This is a testament to the rigorous and highly standardized testing process and safety regulations and procedures for AGT/APM systems (ASCE 21-13; International Electrotechnical Commission [IEC] 62267)12. The AGT/APM industry prepared this important foundation upon which the future mass transit applications of AV driverless roadway technology can build. The functional elements of conventional automated train control systems will be important reference points as new robotic vehicles are deployed in transit service. These aspects of transit operations are defined below from the American Society of Civil Engineers (ASCE) 21-13 APM 11 Observatory of Automated Metros, http://metroautomation.org/. 12 http://orfe.princeton.edu/~alaink/SmartDrivingCars/Stanford_TRB_Conf_July2013/Transit&SharedMobility/Lott_ TRB_Stanford.pdf

TECHNOLOGY 11 Standard and the relevant functionality will be addressed further in subsequent chapters of this working paper. • Automatic Train Control (ATC) – The system for automatically controlling train movement, enforcing train safety, and directing train operations. ATC includes subsystems for automatic train operation (ATO), automatic train protection (ATP) and automatic train supervision (ATS). • ATO – That subsystem within the ATC system which performs any or all the functions of speed regulation, programmed stopping, door and dwell time control, and other functions otherwise assigned to the train operator. • ATP – That subsystem within the ATC system which provides the primary protection for passengers, personnel, and equipment against the hazards of operations13 conducted under automatic control. • ATS – That subsystem within the ATC system which monitors and manages the overall operation of the APM system and provides the interface between the system and the central control operator. 13 Safety analyses of AGT/APM systems identify hazards which describe a condition that could result in an accident, without identifying an accident or potential causes. Distinguishing the hazard from the accident and its causes facilitates hazard analysis and selection of the mitigations designed to eliminate or reduce risk. Examples of hazards are train-to-train collision, train-to-structure collision, train collision with other object, and person struck by a train.

So Singapor followed Source: Kim South Pa urce: Official S Mia e Land Tran by the Circle ley-Horn Figure rk Demons kybus Webpa mi Metromo sport Autho Line (2011 3. 50-Year H tration Proje ge, http://www ver began s rity Northea ) and Down istory of Fully ct – Pittsbur .brooklineconn ervice in 19 st Line bega town Line (2 Automated gh began s ection.com/his 86; Source: K n passenge 013) Guideway Tr ervice in 19 tory/Facts/Sky imley-Horn r service in ansit System TECHNOLO 64 bus.html 2003, s GY 12

TECHNOLOGY 13 Robotic Vehicles in Automated Transit Network Systems – At the time APM/AGT technology began to develop in the 1960s and 1970s, an extension of that concept began to develop for automated guideway systems that would provide a “network” configuration of guideways and stations along which small individual vehicles would operate. The concept included off-line stations such that AVs could bypass on the main line. This birthed the concept of providing “personalized” service directly between a passenger’s origin station to his/her destination station without stopping at any other stations along the route. Originally known as personal rapid transit (PRT) and group rapid transit (GRT), the concept was aggressively pursued through major planning projects and system technology development beginning in the 1970s. One of the first Urban Mass Transit Administration (UMTA) people mover system demonstration projects was the West Virginia University APM system in Morgantown, West Virginia. This system is currently being rehabilitated and remains the only network guideway system in the U.S. with trains dispatched by trip requests of passenger in the stations – a demand-response dispatching concept integral to the PRT/GRT concept14. Over the past 25 years there have been several examples of PRT/GRT systems in small, specialized public transit systems, which deployed robotic vehicles operating along dedicated transitways. These prototype systems (i.e., one of a kind systems) have been operating completely unmanned and steering themselves without physical guidance mechanisms. For purposes of this discussion, each are treated as “guideway” systems in that the vehicles follow a fixed route within a prescribed “transitway” like systems that are physically guided along their path. These robotic systems generally fall into the class of AGT called PRT, GRT, or in more common terminology used in recent years – “automated transit network” (ATN). Three such systems are currently in public transit service using robotic vehicles steering themselves along a fixed-route transitway without physical guidance, although all either calibrate their position from magnetic markers along the guideway or sense the guideway sidewalls using laser technology. Two different size robotic vehicle systems built by a system supplier from the Netherlands are shown in their deployment locations in Figure 4. The larger vehicles in the figure have been operating in a business district since 2005, and the smaller vehicles have been carrying passengers within the office complex since 2010. The other operating system is located at Heathrow airport in London England and began passenger service in 2011. 14 https://en.wikipedia.org/wiki/Automated_guideway_transit

Figure 4 An impor is their o already a with pass elevators operate i the same make ma critically vehicle. T functiona . Robotic Veh tant distinct perational d t the origin engers that works in an n ATNs dem or better le ndatory sto linked to the his on-dem l and opera icle ATN Sys Masda ion of these ispatching “o point when t want to go office build onstrate tha vel of servic ps at each s fact that the and travel fe tional capab tems Curren r City, Abu D Sourc example un n-demand” he passeng that destina ing or hotel t small veh e as single tation. The y do not ha ature is dis ilities of AV tly Operating habi UAE (p e: 2getthere manned rob when passe er arrives, o tion, or com with many f icles can ma fixed-route l cost-benefit ve an emplo cussed furth technology in the Nethe hotos below) otic (self-dr ngers requ ne is dispat pletely emp loors. Dema ke direct or ines utilizing of such aut yee dedica er in Chapt to public tra rlands (photo iving) APM est a trip. If ched to this ty, much like nd-respons igin-to-dest larger vehi omated tran ted to opera er 5 dealing nsit service TECHNOLO s above) and transit syste a vehicle is station eith how a ban e services t ination trips cles/trains t sit systems ting each with applyin . GY 14 in ms not er k of hat with hat is g

Connect CV techn Safety P on a sma technolog with bus congestio Some ap automate demonst PATH) in system in research Toyota a steering achievem (IMTS) a Demonstra ed Vehicle ology evalu ilot in Ann A ll group of t ies on trans crashes hav n after effe plications, s d and conn rated by Ca 2003 using the mid-19 program, a lso demonst during the 2 ent to the in t the 2005 A tion project in and Autom ations have rbor, MI incl ransit vehicl it vehicles a ing significa cts. uch as plato ected comp lifornia’s Par the magnet 90s. The pr nd the deplo rated this ca 005 Expo in ternational SCE Interna 2003 Source Figure 5. ated Bus S been deplo uded CV eq es during th re envision nt severity, oning or co onents. Bus tners for Ad ic nail techn oject was kn yment reac pability of r Aichi, Japa transit comm tional APM : Caltrans VAA Demons ystems – S yed in Ohio, uipment (he e year-long ed to have a typically wit operative-ad platooning vanced Tra ology deve own as the hed full dem unning bus n (see Figu unity as th /ATS Confe Demonstrat tration Projec everal test- Michigan, a re-I-am and evaluation o significant h many inju aptive cruis as shown in nsportation loped for the Vehicle Ass onstration i platoons co re 6). Toyot e Intelligent rence (Aoki ion project in 2 t Source: Ca bed location nd Arizona driver assis f the techno near-term s ries and ma e control, re Figure 5 w Technology automated ist and Auto n 2009. mbined with a presented Multimodal , Hayashida 009 Source ltrans TECHNOLO s for prototy . The USDO tance warn logy. CV afety impac jor traffic quire both as first (Caltrans highway mation (VA automated this importa Transit Sys ). : Caltrans PAT GY 15 pe T ings) t, A) nt tem H

From the past dem purposes over the themselv they mus transit m Lesso Syste Half a ce opportun 1. E of tr a 2. R C ty a b c d Sourc se early dem onstrations (typically w past decade es, tracking t follow to a ay be emerg ns Le ms ntury of exp ities for AV nthusiasm c ten many e ansportation ccomplished esearch and utting edge pically invol . Conceptu technolog . Prototypin . Developm conductin performan design . Commerc e: Wikimedia, Fig onstrations of automate ith a bus op to where v their locatio ny specific d ing althoug arned f erience with transit techn an outpace nthusiasts w , and they p . developme research an ving: al design de ies g major ele ent of test f g numerous ce capabilit ial product/s and: http:/ ure 6. Toyota of unmann d vehicles c erator still s ehicles will s n with preci estination u h there are s rom 50 fully autom ology applic the maturat ho see the ush the dia nt of new te d developm velopment ments and p acilities by p trials to vet ies, and the ystem desig /www.expo200 IMTS Autom ed robotic v onnected w itting in the oon operate se geo-coor sing high-d ignificant c Years ated transit ations in the ion of autom great potent logue to ext chnology re ent of autom of compone roof-of-con ublic and p the system reby find an n with appr 5.or.jp/en/tech ated Buses ehicles stee ith other ve driver’s seat in a truly d dinates, and efinition digi hallenges ah of Aut systems pro next half c ated system ial of autom reme promis quires majo ation techn nts, subsyst cept testing ublic-private concepts, d d resolve w opriate ana nology/imts.ht ring themse hicles for pla ), technolog riverless mo knowing th tal maps. A ead. omated vides seve entury. s’ capabilit ation in the es of what r capital inv ology is ver ems, and in partnership etermine ac eaknesses lyses and fu TECHNOLO ml) lves and fro tooning y has advan de – steerin e specific p new age of Trans ral lessons, ies. There a field of can be estments. y expensive tegrating s and tual in the origina rther testing GY 16 m ced g ath AV it and re , l

TECHNOLOGY 17 e. Producing a complete pilot/demonstration project with safe operations carrying passengers in service conditions suitably representative of the promised deployment operating conditions f. Adapting the design to satisfy transit-grade specifications for system deployment can induce or reveal design flaws that are difficult and expensive to resolve when going beyond the initial proof-of-concept deployments 3. Partnerships are required to deliver a complete operating system. Most transit systems involve integrating many different types of technology and construction, usually including the original technology developer (i.e., the vehicle system supplier), civil and architectural design firms, control and communication system integrators, construction firms, and often financial firms to address interim financing, bonding, and insurance requirements. Hurdles to Deployment – Bringing advanced technology to the market place can face hurdles not apparent when the pursuit began. Several aspects of slowing deployment have been seen in the historical context of advanced transit system technology, such as: 1. Funding regulations constrain sources of Automated Transit supply – Transit system procurements within the U.S. which are made with grants from the FTA require a significant percentage of the system supply, including the transit vehicles, come from U.S. sources of supply. There are a variety of ways to satisfy the specifics of the “Buy America Act” requirements, but tracking and documenting all supply sources for system equipment and fixed facilities is a major hurdle that must be cleared for any federally funded transit project. 2. Labor agreements may constrain fully automated operations – Even when full automation is accomplished and there is no longer a need for operators or attendants onboard, there have been circumstances where labor collective bargaining agreements required a human operator be retained at the front of each train or onboard as an attendant. 3. Owner/operator transit agencies want someone to backstop their risk – Any new technology applied early in its development cycle requires a sharing of risk – both with respect to operating costs and liability. 4. Acceptance by industry professionals and system operators will take time – The acceptance of fully automated transit in the U.S. has been slow to take root. For most of the past 50 years, fully automated systems have only been deployed within or connecting to major airports, but have been rejected as a technology of choice for more conventional transit lines. 5. Public acceptance of automated systems takes time – The uneasiness that a passenger can feel when riding in a vehicle with no human operator can be a factor in the public acceptance of the fully automated transit system. And major publicity over any crash or collision involving an AV vehicle of any type will add to this discomfort. 6. Challenges of “Safe” system design require extensive analysis and testing – The transit industry’s venture into full automation has historically been based on rigorous safety analyses that have been derived from the aerospace/military industry. 7. Americans with Disabilities Act (ADA) mandates for Transit Systems are particularly difficult for fully automated systems – Of major importance with respect

TECHNOLOGY 18 to fully automated transit systems is the U.S. government’s enforcement of the Americans with Disabilities Act15. This set of governmental regulations has specific requirements for many aspects of a disabled transit patron’s ability to access public transit. Implications of Technology Readiness for Transit Caltrans PATH work dating back to 2003 demonstrated several automated transit functions including automated platooning and automated steering on transit vehicles16. These prototype AV technology features were demonstrated in revenue service in Lane County, Oregon in a pilot deployment. The approach uses magnetic nails/markers embedded in the pavement along the bus rapid transit (BRT) route (and did so as well in the freeway tests in 2003). Sensors – Passenger vehicle technologies, spurred by the Defense Advanced Research Projects Agency (DARPA) challenges in 2005-2009 have outpaced developments specifically targeted for general purpose transit operations. The integration of enabling technologies (sensors) for automated operation is just beginning to find its way into buses. As sensor technologies continue to advance, transit specific versions will need to address bus-specific form factors for equipment installation, but also sensitivities, placement, field of view, and other parameters different for modern transit vehicles than for passenger vehicles. Artificial Intelligence – Artificial intelligence algorithms also need further development specifically for transit applications. Buses do not respond the same as passenger vehicles to basic throttle and steering commands and have more challenging requirements for algorithms that merge a bus into a stream of vehicles, weave across several lanes of traffic, or execute left- turns in intersections, for example. It is not enough to just drop in an algorithm(s) designed for passenger vehicles (or trucks, for that matter) and automate transit vehicle operation. Significant work is necessary to modify the operating parameters of these methods for transit operation in general mixed traffic environments. The low-speed shuttle mode operating at Level 4, however, has shown significant developments over the last five years with several systems in revenue service and a host of new pilot deployments coming online across the world. These vehicles work with existing guidance and control technologies by substantially simplifying the operating environment (and thus the complexity of their control algorithms) and lowering the speed to minimize the severity of failures. The significant opportunity for automation in transit is likely scaling up the technologies developed for low-speed shuttle operation to use on common bus form factors Communications – All shuttle operations require significant bandwidth and continuous communication links for remote monitoring and piloting in the event of vehicle system failures. Existing communications methods should be adequate for such operations and not impede any development of automation in transit services. There is no debate that CV technologies which link vehicles to continuous data streams the roadway infrastructure and from other vehicles about the operating environment can substantially enhance automated operations. There is 15 https://www.access-board.gov/guidelines-and-standards/transportation/facilities/ada-standards-for- transportation-facilities 16 http://www.path.berkeley.edu/sites/default/files/publications/PRR-2009-12.pdf

TECHNOLOGY 19 nothing inherently precluding computers from ingesting data from existing CV concepts and acting on it automatically, except that in many cases some information transmitted is expected to be consumed by a human driver (particularly the general category of “traveler information”). Mapping – High-resolution maps of the roadway network and street infrastructure are critical for enabling generalized Level 3 and Level 4 operation of AVs including transit vehicles. HERE has notably identified this as a market need and is beginning to offer this as a service17. Road centerline maps enabling route guidance for human-driven vehicles simply cannot be used by AVs for tactical negotiation of the roadway environment. Onboard storage of such a sub-lane- level precise (and hopefully accurate) dataset is formidable and requires more than what can be easily stored on a $99 navigation device. The data regarding traffic control elements such as speed limits, stop signs, traffic signals, turn restrictions, and so on is a similar need for Level 3 and 4 operation and specific data relative to transit operations will be needed for general operation of AV Transit. Since these data are typically managed by a myriad of state, county, and local government entities today, a standardized database of the infrastructure assets will almost assuredly need to be managed by a third party(s) or the federal government. This is a formidable challenge to generalize operation of AVs at Level 3 and Level 4 across the U.S. and the world. AV Enabling Technologies and Transit Applications Service applications of automation technology within transit vehicles are an important first definition of AV introduction to public transit. Subsequent chapters will address the associated facilities and operational considerations of AV introduction into transit systems. Table 1 illustrates the correlation of human/machine interface functionality, and transit vehicle capabilities with progressively higher levels of automation on the NHTSA/Society of Automotive Engineers (SAE) scale. Table 2 is organized as follows: 1. NHTSA/SAE Automation Level provides a first level correlation to the AV enabling technologies matrix presented in Table 2. 2. HMI Operational Classification Level provides a basic description of the human/machine interface (HMI) in each transit vehicle as the AV functions move from Level to Level. This indicates the level of responsibility, skill, and attention a human must maintain as the transit vehicle operates within its given operating environment. 3. Example Automated Machine Functions indicates a correlation to the AV enabling technology matrix of Table 2. Note that these are examples, since a comprehensive description is beyond the intent of this summary. 4. Transit System Applications provides a representative explanation of transit system application, without attempting to provide a comprehensive discussion. 5. Potential Deployment Timeline. Although timelines for deployment are difficult to forecast, the times indicated are a first attempt at assessing when a mature functionality for general transit use will be possible. 17 https://company.here.com/automotive/intelligent-car/here-hd-live-map/

TECHNOLOGY 20 Note that the timeline for technical feasibility does not consider the separate timelines for institutional changes to operating policy, governmental agency regulations, and associated laws. These aspects will be addressed in subsequent working papers. Table 1. Human/Machine Interface Functionality NHTSA/SAE Automation Level HMI Operational Classification Level Example Automated Machine Functions Transit System Applications Potential Deployment Time Line 0. No Automation Human driving None Conventional roadway transit vehicles, no automation Today Human driving with warnings Forward Collision Warning (FCW), Blind Spot Warning (BSW), Lane Departure Warning (LDW) Conventional roadway transit vehicles with necessary sensors that provide warnings now and may enable automation later Today - 2020 1. Function Specific Automation Human driving with machine assistance Adaptive cruise control, lane following, emergency braking (separately) Safety-enhanced conventional roadway transit vehicle a.) Enhanced technology buses 2015-2020 b.) Enhanced technology automobiles (e.g., ride-share vans) 2015-2020 2. Combined Function Automation Machine-driving in special environments for enhanced safety Adaptive cruise control, automated braking, and lane following (together) Advanced technology roadway vehicles with platooning with an operator on each vehicle monitoring the automated driving functions a.) Special environment: buses in High occupancy vehicle (HOV)/managed lanes 2020-2025 b.) Special environment: BRT in exclusive transitways with controlled at-grade crossings of city streets and pedestrianways 2015-2025

TECHNOLOGY 21 NHTSA/SAE Automation Level HMI Operational Classification Level Example Automated Machine Functions Transit System Applications Potential Deployment Time Line 3 Limited Self- Driving Automation Machine-driving with human oversight Automated driving over portions of a route with substantive travel distances, but with human operator available to take control if required Automated operations between stations; onboard attendant (present for failure management and emergency incident management) a.) Special environment: buses in HOV/managed lanes 2020-2030 b.) Special environment: BRT in exclusive transitways with controlled at-grade crossings of city streets and pedestrianways 2020-2030 c.) Mixed traffic environment: local bus routes and demand-response dispatch service on local city streets and arterials 2025-2035 Automated driving with high-precision maneuvering at low speeds Automated operations during high-precision maneuvers; onboard attendant (present for failure management and emergency incident management) a.) Station approach and docking maneuvers at platform berth 2015-2020 b.) Precision maneuvering in storage areas or within maintenance depot 2015-2025

TECHNOLOGY 22 NHTSA/SAE Automation Level HMI Operational Classification Level Example Automated Machine Functions Transit System Applications Potential Deployment Time Line 4. Full Self- Driving Automation Machine-driving without human presence required; provisions for human-driving operations by roving “recovery” personnel or by remote control from a centralized or nearby location Automated driving, path determination and station berthing without a driver onboard at any time from origin to destination, Automated transit route or demand-responsive dispatch operations; empty vehicle repositioning/storage a.) Special environment: protected (e.g., campus) environment on dedicated transitways at low operating speeds 2015-2020 b.) Special environment: automated HOV/managed lanes with operator boarding at HOV/managed lane facility exit station stop 2025-2035 c.) Special environment: BRT in exclusive transitways with controlled at-grade crossings of city streets and pedestrianways 2025-2035 d.) Mixed traffic operations (i.e. interacting with other non-automated vehicles) at low speeds on city streets 2025-2035 e.) Mixed traffic operations (i.e., interacting with other non-automated vehicles) at all speeds and in any roadway operating environment 2030-2050

TECHNOLOGY 23 Potential Evolution of New Transit Paradigms The conventional transit bus coach has evolved to the 40’ bus size typically used today because it provides a good balance of cost-benefit when the bus is full (driver compensation, fuel, and other operating costs offset by transit fares). Similarly, the use of 50’ to 80’ rail cars has provided the backbone of transit service in high-demand travel corridors over the past century. But inefficiencies of many current transit systems result simply because the buses and trains are not full of riders on the route throughout the day. From the passenger’s perspective, in many communities it simply takes too long to get from an origin to a destination as the transit vehicle/train makes many stops along the route, and transfers between one route and another add additional waiting time. Further detrimental impacts to transit ridership are created when transit agencies invest in expensive line-haul systems on major routes with the objective of raising the benefit-cost ratio for transit, while creating the last-mile/first-mile connection problem in doing so. If transit vehicles can be made smaller and be deployed to operate in more of a point-to-point type service on roadways using demand-responsive automation like an ATN, we believe that trip times of individual patrons will likely become closer to private autos or taxis, bringing more transit users to the system. By removing the overhead cost of having an operator on every vehicle or train from the cost of fleet operations through automation, we believe the benefit-cost ratio of such an AV-based system could become an attractive option for transit agency investment, with the added stipulation that the regulatory and operational issues are addressed. Near- and Medium-Term Operations We posit that the earliest applications of AV technology to transit will involve the operation of buses as they travel along dedicated transitways such as exclusive BRT corridors, within HOV roadway facilities, or on existing bus on shoulder routes. In the near term, these facilities can be upgraded to allow AV technology to autonomously steer the vehicles, perform propulsion and braking control, operate in multi-bus platoons, and provide collision avoidance protection. AV technology will allow the BRT vehicles to be platooned (or “virtually coupled”) to create more train-like operation without the need for the track of light rail transit (LRT), and likely at reduced cost with similar line-haul capacity. This concept of dynamically reconfiguring a train of AVs is also being pursued for commercial trucks in the U.S. and Europe with serious emphasis on near-term operation due to cost savings due to fuel efficiency 18. Anti-platooning and close- following laws in several states are critical regulations that need to be addressed (not only for trucks, but for buses in BRT lines) and a new research project addressing these legal constraints is needed. Another development expected in the near term is the blending of the previously developed and demonstrated guidance technology using magnetic markers with rapidly advancing high accuracy global positioning system (GPS) technology, Inertial Measurement Units (IMUs) and high-definition maps. The combination of these technologies allows the vehicles to operate in a 18 https://www.eutruckplatooning.com/About/default.aspx

free-rang the imme and route We belie like settin relatively campuse operating low risk o CitiMobil mixed tra As collisi continue situations These ev Long Operatin developm respond not most As an ex systems BRT line 19 http://ww ing mode a diate vicinit alignment. ve the seco gs where th slow speed s, large med at reasona f injury or c 2 project co ffic and bas on avoidanc to evolve an , speeds ca olutionary s -Term g concepts m ent of AV t dynamically fixed-route ample of thi to multiple p . Conceptua w.citymobil2 long some o y of station s Figure 7. Toy nd context fo ere is a sem s. This envi ical comple bly low spee rashes. This nducted by t ic traffic sig e, object/pe d become m n be increa teps will pro Operat ay pass th ransit applic to changing transit opera s paradigm hysically or lly, the futur .eu/en/ f the route, tops using ota IMTS Au r early tran i-controlled ronment is c xes, and m ds can inte scale of de he Europea nals as well rson detect ore accura sed and a w bably take m ions rough a com ations. In fa demand pa tions. shift is the c virtually ent e world of A but then usi to provide e tomated Bus sit deployme environme ommonly fo aster-planne ract with oth ployment ha n Union, inc , at very low ion technolo te and capa ider range o ore than 1 plete parad ct, the flexib tterns will li onceptual c rained AV r V transit tec ng the magn xtremely ac es Source: W nt of L4 AV nt in which v und in colle d communi er traffic, pe s already b luding limite speeds19. gy, and auto ble of hand f roadway f 0 years to b igm shift du ility of future kely gradua onversion o ubber-tire ro hnology co etic marker curate and r ikimedia transit will ehicles can ge and univ ties where a destrians, a een accomp d operation mation con ling more co acilities can e safely imp ring the long transit sys lly begin to f fixed guide adway trans uld allow the TECHNOLO technology eliable dock be in campu operate at ersity transit veh nd cyclists lished in th s on roads w trol algorith mplex be navigate lemented. -term tems that replace man way LRT it vehicles a same vehi GY 24 in ing s- icle at e ith ms d. y, if s a cles

TECHNOLOGY 25 to operate on some combination of dedicated transitways and/or conventional roadways while operating along their assigned travel path. In this potential future, a fleet of smaller automated transit vehicles could also be dynamically repositioned through strategic distribution anywhere in the transit network to serve changing demand patterns. It is likely that the typical transit services during busy times of the day will include multiple riders bound for the same destination from the same origin or with a limited number of stops for pick- up or drop-off on a common route. We believe the difference from typical fixed-route, line-haul transit operations today will be the more direct origin-to-destination station service with fewer stops along the travel path of every transit patron. Transit users will have a travel time that closely matches the personal automobile. Empty Vehicle Management is especially important since during significant periods of time (e.g., at night) there will typically be many fewer trip requests than during the peak periods. During those off-peak times the automated supervisory control system will send empty vehicles into storage locations placed throughout the transit network, typically near the portions of the transit network where high demands will arise during the next peak activity period. Then as trip requests are received, the supervisory system will dispatch a nearby and available empty vehicle to pick up the transit patron(s). It is this functionality that optimizes the use of energy and vehicle-miles by automatically removing vehicles from service as ridership demand drops. Potential Near- and Medium-Term Changes to Fixed Facilities The changes to transit facilities in the near term necessary to accommodate the new operational concepts for Levels 3 and 4 automated vehicles will begin to impact both conventional roadway and guideway transit facilities. The challenges of full automation and facilities that comply with safety and ADA requirements will be particularly challenging, particularly for locations that are planned to be built in the next 10 years. Transit Stop Locations – In general, as transit operations begin to employ on-demand features that allow the transit users to create more customized trips, the number of locations where transit service can be accessed could increase. This also brings consideration of an increase to the number of weather-protected shelters at new transit stops. With the origin/destination options increasing from what is provided today with conventional on- demand public transit service, the realities may include the need to provide enlarged zones for public transit vehicles to stop adjacent to or within high-demand trip generators like urban districts, university campuses, high-capacity rail stations and airports. Adequate provisions for protected boarding and alighting need to be provided for the number of large and small transit vehicles that may arrive during high activity periods. Transition Zones at Protected AV Operating Environments – Early applications of Level 3 automated driving within protected environments like HOV/managed lanes of shoulder lanes will necessarily require the transit vehicles to transition back to primarily manual control as the vehicles leave the protected areas and enter mixed traffic operations still with perhaps Level 1 or Level 2 features available to the driver. These transition zones may eventually be possible while the vehicle is moving at high speeds, but for the near term the provision of a transition

TECHNOLOGY 26 zone where the vehicle can be brought to a stop or substantially reduce the operating speed is likely to be necessary while the transition occurs. Multi-Berth and Off-Line Stations – Transit station facilities in the near term will begin to change from fixed guideway station configurations. Starting in the near term, the functional ability to platoon AVs will immediately require BRT stations to accommodate multiple vehicles simultaneously stopping in each station along the line. This is the most eminent functional capability that will impact the conventional configurations of existing bus rapid transit facilities. In the intermediate term, conventional online stations, at which all vehicles/trains passing along the main line transitway must stop at every station to allow any passengers to board or alight each vehicle/train, will likely gradually be replaced by off-line stations. Many vehicles (or virtual trains of vehicles) will bypass many stations without stopping since the transit supervisory system (e.g., the fare collection and vehicle dispatch system) will know if any passengers need to board or alight at each stop. Maintenance and Storage Facilities – Maintenance facilities for AV automated transit systems will be configured much like conventional bus maintenance facilities, whereas the storage facilities can be located anywhere that is accessible to the route. Storage areas placed in locations away from the maintenance facility will be dynamically utilized throughout the day. Each storage facility’s strategic placement and capacity will be designed to hold a portion of the operating fleet in a “hot standby” mode, until such time each vehicle is dispatched back into passenger service. There will still need to be storage in or near the maintenance facility, since each vehicle will need a pre-service checkout and test, as automated guideway transit systems go through today. However, remote diagnostic checkout of all functions of vehicles will likely be possible due to existing wireless communications and software technology (e.g., Tesla vehicles and many other OEM vehicles get software updates over the air in 201620), thus eliminating the need to size the maintenance facility storage areas to hold the whole operating fleet. Findings on AV Technology Deployment in Transit Service AV technology will impact the public transit industry in a dramatic way during the next two to three decades. Transit service types (fixed-route, demand-response, etc.) will be the key determinant of the business models by which transit services will be delivered. AGT/APM maturation over the last 50 years has shown that design of transit systems with automated functions must be applied in an integrated fashion across multiple subsystems (e.g., vehicle driving, vehicle location determination and guidance, vehicle/station berth interface, V2V and V2I communications, etc.). Enabling AV transit technology is by its nature a complex system but is maturing rapidly. Technology is not expected to be the limiting factor for transit applications, unless the safety requirements are made so stringent that systems are too costly or too complicated to deploy. 20 http://arstechnica.com/cars/2016/01/finally-over-the-air-software-updates-for-your-car-are-becoming-a- reality/

TECHNOLOGY 27 Specific designs for large transit vehicles combined with progressive demonstration in test environments will likely be the path toward improving safety and mobility of transit operations through automation. Timelines expected for AV transit technology readiness are: • Near term (5-10 years) will see applications of AV transit technology to BRT transitways and HOV lanes, in addition to more advanced technology applications for L4 vehicle location determination, guidance and pathing in controlled environments such as campuses. • Medium-term (10-15 years) will reach L4 driverless vehicle operations in HOV, BRT, and low-speed mixed traffic environments. • Long-term (15-30+ years) will have AV transit vehicles operating in all environments and will be integrated into fully automated transit systems. Subsequent working papers explore in more detail the issues and barriers to adoption of AV transit technology by transit operating agencies. These considerations will frame the roadmap of activities needed to overcome these barriers and improve safety and mobility for transit patrons through automation. Research Projects on AV Technology Deployment in Transit Service – The timeline for initial deployment of AV technology in transit service starts now, and the early years of partial automation will be as important as the later years of full automation. The key research projects for undertaking based on the considerations and findings of this working paper are as follows: 1. AV Transit Liability, Insurance and Risk Acceptance – Research would be helpful on the liability aspects and insurance coverage that will be distributed between the vehicle manufacturer, the operating agency and the human operator, particularly for times when transitions from automated vehicle control to human operator control is a frequent occurrence. The area of focus would be from legal and contractual (collective bargaining) considerations. The related aspects of employee and passenger “acceptance of risk” when onboard public transit vehicle where the human operator is no longer responsible for all functions required to operate the vehicle is a related area also needs further legal research, which could be addressed under this project. 2. Legal Constraints to Platooning and Virtual Coupling – The concept of dynamically reconfiguring a train (platoon) of AV vehicles has relevant application both in the near term and increasingly in the medium and long term. Anti-platooning and close-following laws in several states are critical regulations for this research project to determine their legal application to buses in BRT transitways or HOV lanes for the near term. And for the long term the legal implications of such laws on lower-speed arterial street as well as high-speed freeway operating conditions would be beneficial if considered. 3. Features and Configurations of Transit Fixed Facilities – Beginning with an assessment of the practical and technical implications for providing more direct service without intermediate stops using off-line stations, a beneficial research project would evaluate the implications for operations in line-haul high-capacity. The work would evaluate how this new concept could potentially allow almost all stations to be designed for fewer number of vehicle berths. Near- to medium-term changes to transit fixed facilities research activities would include exploring features and right-of-way requirements for station/stop locations, transition zones from the main operating lanes

TECHNOLOGY 28 into off-line stations, and the configuration of multi-berth boarding positions. In addition, precision docking can enable all stops to offer level boarding for the physically challenged. 4. Virtual Entrainment of AV Transit Vehicles – Additional research would focus on the long-term implications of dynamic entrainment with virtual coupling/uncoupling to allow longer “trains” moving through the transitway/roadway system then separating into individual vehicles when berthing at stations.