Cooperative Intersection Collision
Avoidance Systems (CICAS)

CICAS Workshop Report

December 9–10, 2004
Churchill Hotel
Washington, D.C.

Summary of CICAS Workshop Activities and Results

The CICAS Workshop activities included the following:

• Stakeholders from the FHWA, NHTSA, FTA, FMCSA, automobile industry, state DOTs, and university ITS research groups convened to present their various approaches to CICAS.
• The Crash Avoidance Metrics Partnership (CAMP), a research consortium of automobile manufacturers, presented its vision for CICAS. CAMP described in detail a model for potential solutions to intersection collision avoidance systems and the evolutionary path from a cooperative vehicle-infrastructure system in the near-term to a future system of autonomous vehicles communicating with each other about impending crashes.
• The FHWA presented summaries of its current research projects, and of the research being conducted by the States and their university partners.
• The USDOT presented its vision for CICAS, including the requirements for the output of the program at its scheduled end date of 2009.

The workshop resulted in:

• A detailed discussion of both the USDOT's requirements for CICAS and a debate about the CAMP model from the perspective of the other stakeholders
• Acceptance of the high-level requirements for a cooperative system as outlined by the USDOT and by a diverse set of stakeholders
• An outline of the research agenda that will be needed to reach the USDOT's goals for CICAS
• A charge to the States and the Infrastructure Consortium to build a model of the solution from the point of view of the infrastructure. This model will eventually merge with the CAMP model and will encompass the USDOT's requirements for CICAS
• The beginnings of a research path to achieve a field-tested version of CICAS by 2009
• A beginning of a vision of what CICAS should be in 2009
• A suggestion to reconvene in January with representatives of additional stakeholder groups, e.g., traffic device manufacturers

Day One – 12/09/2004

Welcome
Mike Schagrin, ITS JPO
Mike Schagrin welcomed the group to the Cooperative Intersection Collision Avoidance Systems (CICAS) Workshop and provided an overview of the workshop including the need for a consensus on the vision, development of a high-level roadmap, a better understanding of how to invest funds for program activities and support.

Crash Avoidance Metrics Partnership (CAMP)
Vehicle Safety Communications Consortium
IVI Light Vehicle Enabling Research Program
Task 15

Rajiv Gupta, GM
Michael Maile, Daimler Chrysler
Ron Heft, Nissan

The Crash Avoidance Metrics Partnership (CAMP) provided an overview of their work to date with regard to Cooperative Intersection Collision Avoidance Systems (CICAS). CAMP has been working to develop a Concept of Operations (ConOps), requirements, specifications, and the design of a driver-vehicle interface (DVI) for CICAS. They provided a review of existing technologies, including dedicated short-range communications (DSRC). The scope of CAMP's ICA effort led them to develop seven models or concepts of how ICA could be implemented, ranging from infrastructure only to vehicle only options, with combinations of the options as intermediate positions. Details of the concepts can be found in the report: Vehicle Safety Communications Project Task 15: Preparatory Analysis to Support Intersection Collision Avoidance (ICA) Workshop Discussions, CAMP, November 19, 2004. See workshop slides for outline of overview.

CAMP, by definition, presented a vehicle-centric view of the evolutionary path to a mature, operational CICAS. However, the CAMP model does include infrastructure-centric components that provide data communications capability to DSRC-equipped vehicles and driver interface capabilities to non-DSRC-equipped vehicles. Key elements of the CAMP model include the following:

• DSRC is the enabling communication technology with a roadside unit (RSU) and an onboard unit (OBU) always present.
• The location of sensing, computation and/or warning in the vehicle or infrastructure depends on the model concept being implemented in the vehicle.
• The driver-vehicle interface (DVI) alerts the driver and is functionally integrated with other safety systems.

Table 1 indicates the concepts that form the CAMP solution. Table 2[1] arrays the concepts along several descriptive dimensions.

Table 1: CAMP CICAS Concepts

	Scenario	Description
1	Infrastructure Only; No DSRC	Infrastructure autonomous system
2	Infrastructure Decision, Vehicle Display; DSRC	Starting point for Infrastructure decision based CICAS.I2V communication required for in-vehicle display
3	Infrastructure Supported, Vehicle Decision	Starting point for Vehicle decision based CICAS
4	Vehicle Supported Infrastructure Intensive Infrastructure Decision	Vehicles all send their information to infrastructure. Infrastructure does calculations and issues commands to particular vehicle to issue warnings.
5	Vehicle and Infrastructure Supported, Vehicle Decision	Vehicles all send their information to the infrastructure. Infrastructure compiles information and provides intersection 'dynamic state' to all vehicles (DSRC equipped and non-equipped). Vehicles warn as appropriate.
6	High DSRC-Vehicle Penetration, Intersection Supported, Vehicle decision	All vehicles transmit their relevant information. Infrastructure re-transmits vehicle messages and any infrastructure information. Vehicles receive messages from all other vehicles and warn appropriately
7	Vehicle Only	Vehicle autonomous systems. Based entirely on V2V communication

Table 2: CAMP CICAS Concepts - Descriptive Array

		Data Flow Direction			Sensors		Threat Assessment		Interface		Problem Area
Con-cept #	Title	I2V	V2I	V2V	I	V	I	V	DII	DVI	Violations	Gap
1	Infrastructure only; No DSRC
2	Infrastructure decision; DSRC
3	Infrastructure data; vehicle decision; DSRC
4	Vehicle data; infrastructure decision; DSRC
5	Mixed DSRC/Non-DSRC
6	High DSRC
7	~100% DSRC

CAMP proposes that CICAS development start with Concept 3, proceed to the intermediate system architecture suggested by Concept 5, and extend to Concept 6 as DSRC-equipped vehicles become more common in the vehicle fleet. Concept 7 would be the final, long-term result of CICAS development. CAMP also recommended the formation of a working team of infrastructure and vehicle experts to develop the architecture, communications system, and deployment systems.

Much of the discussion involved reconciling the vehicle-centric view with an infrastructure-centric view. The following questions and issues emerged from the group discussion following the CAMP presentation:

• Pedestrians: The positions of pedestrians, bicyclists, dogs, horses (or generalized traffic participants) were not present in the state map of the intersection described by CAMP; CAMP responded that this capability belongs in Concept 5, and that Concept 6 assumes that there will be enough additional sensors on equipped cars to sense pedestrians and unequipped cars. Pedestrian sensing is a general problem that needs to be addressed in future program activities. Decisions are required for how pedestrian sensing will be performed (infrastructure or vehicle) and how the warnings will be delivered.
• Warnings: All warnings are in the vehicle. CAMP Concept 3 does not exclude infrastructure-delivered warnings. However, the infrastructure perspective was not in CAMP's scope. Some workshop participants believe that infrastructure communications/warnings may be a better alternative than in-vehicle ones, because infrastructure warnings can warn all vehicles at the same time while in-vehicle warnings only warn equipped vehicles. These participants believe that in-vehicle warnings may be suboptimal at this time.
• Relationship of Concepts: Combining Concept 1 and Concept 3 is, in effect, pre-Concept 5.
• Concept 3 Benefits: The benefit is proportional to the number of DSRC-equipped vehicles. The States felt that if too few vehicles were equipped, there would not be enough benefit to justify the cost of the infrastructure component to the States. CAMP responded that it is recommending that infrastructure only development be conducted in parallel to its development efforts.
• Latency: The controller-RSU interface has a specification of 10 Hz and 100 ms. CAMP feels that this is sufficient for the application, but CAMP has not yet worked out the overall system latency budget.
• Harmonization of Infrastructure and In-Vehicle Warnings: Participants suggested that this is an issue on which further research is needed.
• Combined Error: The combined error from the map of intersection geometry plus the error from GPS is 10m, which translates to a one second error. Is this enough accuracy? What is the accuracy needed for the state map? CAMP does not have a good fix on the requirements, particularly for lane level positioning where 1m accuracy is needed.
• Gap Acceptance: There are, in fact, two types of gap acceptance depending on whether the vehicle is stopped or moving. CAMP has not decided whether its gap acceptance research will be focused on stationary or moving vehicles. There was some discussion as to whether drivers time the gap or gauge distance; a large body of research exists on this type of driver behavior.
• The Dilemma Zone: A participant from the FHWA stated that the signal state is more complex than portrayed, including instances of an extended green phase and a variable time for the yellow. Traffic behavior in the dilemma zone needs to be studied. If a vehicle stops at the yellow in a high-speed intersection, this is likely to cause rear end collisions. CAMP answered that in the future they will need to generate requirements for traffic controllers, so that CICAS can anticipate signal change rather than just react to it. It was recommended that CAMP work with State DOT researchers to better understand needs and develop requirements. There was also discussion of acceptable deceleration rates (.3g) for autos in the dilemma zone.
• DII: A comment was that in Concept 5/6 the risk level for the DII should be changed to "red" as DII represents a high-risk area requiring further research.
• Starting with Concept 2: A participant suggested that the evolutionary path should start with Concept 2. CAMP picked its start point as Concept 3, because it knows the most about vehicle-based systems. In addition, in Concept 2, the system may not have enough information available, because the system does not know the braking capability of the vehicle, or whether the driver's foot is on the brake. DSRC and in-vehicle messages are going to be used for many different applications
• Rural/Urban/Suburban Suitability: A question was raised that the CAMP concept was unsuitable for rural areas. CAMP representatives debated this point and provided evidence for why they felt differently.
• Relationship Between Infrastructure-only and Vehicle-based Systems: Issues raised included:
• The understanding that the infrastructure delivered a message to the driver, not a warning
• The distribution of liability among automobile manufacturers, device manufacturers, and public transportation agencies in the various concepts
• Whether Concept 3 is a feasible starting point without going through Concept 1 or Concept 2
• The ability (or lack of ability, as stated by VTTI) of vehicle-based systems to detect a car behind a truck
• The low likelihood that the States will invest in a system that requires a high penetration of DSRC
• The evidence, if any, that in-vehicle warnings are better than infrastructure warnings. (Which leads to more effective driver action? CAMP stated that there are fewer false alarms with in-vehicle warnings.) One participant questioned whether, since about 50% of intersection violators are willful violators, will a DII shame them into following the rules?
• Partial or Full Automation: Is either a better solution than providing decision support to the driver? Given the level of uncertainty of the data can we move to automated control? Full discussion was deferred to the breakout session.
• Concept 5/6: What is the decision point to move from Concept 3 to Concept 5/6? The suggestion was that there will be a necessity to do Concept 5 or maybe Concept 4 to have a combined infrastructure/vehicle solution. Another suggestion was to more fully incorporate ranging sensors into the vision for Concept 5/6. CAMP answered that the cost of sensors is prohibitive at this point.
• Willful Violators: What about willful violators? (These were 40% of the intersection collision problem 10 years ago.) DII may actually change driver behavior, whereas a violator can ignore an in-vehicle warning.
• Policy and Business Model Questions: What business, voluntary, and regulatory deployment models are needed? Deployment of the needed equipment into new vehicles will not affect the existing fleet. It takes 7 - 10 years to turn over the fleet, which is under the control of state licensing and inspection agencies. There was discussion about the likelihood of success of DSRC and its penetration into the market possibly being another hurdle to the deployment of CICAS. There was discussion of the fact that, even if 100% of new vehicles were equipped with DSRC, it would take a number of years for such vehicles to become a large percentage of the US vehicle fleet.
• Testing and Implementation: VTTI has concluded that implementation is not straightforward. There are too many type 1 and type 2 errors in their tests. NHTSA noted that there will be a 100-car study with fully sensor-equipped cars; CAMP will follow up. Another research effort will be looking at intersections for next 4-5 years.
• Driver Performance Research: There seems to be a significant hole in driver performance research. Where does the research gap lie, with CAMP or other research efforts? When will this research be done and by whom? Will new and existing information be gathered by FOTs and OEMs?

Current USDOT related research projects
Gene McHale, FHWA, R&D

Following McHale's presentation on current FHWA research projects, the discussion centered on research needs and the kinds of data that are being collected. VTTI listed ideas for future research:

• Collect live data at intersections – for example, such as determining the effects of infrastructure-based sign near Blacksburg
• Examine in-vehicle and infrastructure warning interaction
• Mine existing database sponsored by NHTSA, which contains naturalistic data for 100 cars driven throughout controlled intersections in northern Virginia. The data collection will end in September.

CAMP has eye-tracking data (which includes 1200 approaches to intersections). In addition, participants noted an evolutionary approach that could start at Concept 1, and then proceeds through Concept 3 to Concept 5. The state map is being examined, as some entity in the system must have a picture of what is happening at an intersection.

State DOT Perspective
Morris Oliver, FHWA, Office of Safety

Following Oliver's presentation there was extensive discussion between the States and CAMP, and with Jim Wright, the AASHTO representative. CAMP asked: How do the States see themselves in getting organized in dealing with CICAS? Are we talking about all 50 States or just the five involved with IDS?

AASHTO replied that the States have much experience in working together, e.g., VII and DSRC. There are no specifics yet with respect to CICAS. AASHTO is working on this issue with the CICAS Program Manager, Mike Schagrin, and will have an answer in next few months. The intent is that there will be one single point of contact for all of the States.

Schagrin says that he intends to use a model similar to CAMP to bring the States together. One unified, endorsed position will come out of the model and USDOT will facilitate discussions among the States to come to a consensus position.

CALTRANS mentioned two additional challenges:

• Supporting legacy systems: Signal controllers have been deployed for decades and it is unlikely that they will be replaced en masse to support CICAS infrastructure needs
• Local governments: Most intersections are owned and operated by local governments, not States, although CALTRANS issues standard plans and specifications to local governments. Local agencies need a voice in CICAS.

States present at the workshop emphasized that the intersection is only one component of VII. Other components are important to DOT as well, not just this component.

AASHTO is making a commitment to educate and spread the word about the value opportunities of these programs. What will be the design-build-deployment process? What is the overall business model? AASHTO will work at the executive level with the automobile industry and USDOT.

VDOT stated the state's case for a key role for infrastructure-based systems. The state DOTs have a daunting task in terms of safety and they have long history of working together. The state perspective agrees with CAMP on the need for an incremental rollout. However, the infrastructure has to be the first phase (in the near term), and has to be a significant part of the intermediate solutions. This is because the States cannot wait for the market penetration of DSRC. The "near term" is defined as the take-up time for DSRC and the deployment of I2V systems.

VTTI stated that there is a need for a common vision of an evolutionary deployment. An infrastructure-only system to start gives car companies a foundation against which to design their vehicle-based systems. The States will see immediate safety benefits from the infrastructure-based systems, because they will be deployed at the most problematic intersections. However, the maximum benefit will not be attained, until there is a cooperative system, which will need to be recognized in the requirements for earlier implementation phases.

A consensus emerged among the States about developing an infrastructure solution that merges with other solutions sometime in the future. In the near term, before DSRC-equipped vehicles are a significant portion of the U.S. fleet, States could focus on infrastructure deployment of major components. The challenge is to work closely together. Infrastructure is a significant part of the incremental solution and States may not be able to afford to wait for DSRC-equipped vehicles. The infrastructure approach is critical to getting to cooperative intersection collision avoidance systems. The progression could be as follows:

• Build Concept 1
• Develop Concept 3
• Hold Concept 5 as a vision

Further discussion included the following points:

• Infrastructure vs. in-vehicle warning conflict: this is a research issue; it may be that both together are beneficial. One alternative is that the infrastructure generates the warning for both the DII and the DVI, so that there will be no conflict.
• Benefits: From a public sector point of view, there needs to be an immediate benefit. Otherwise, the States won't pay for the capability.
• Human factors: Private sector data on user acceptance is needed.
• DSRC transmitter deployment: The VII group is addressing deployment.
• What is the difference between "signal violation," which CAMP's Concept 3 addresses, and "gap assistance," which CAMP addresses in its Concepts 5 and 6? Signal violation countermeasures provide a warning about a potential collision and gap assistance countermeasures help a person to make a good decision.
• Why are rear-end collision warnings not being addressed by CICAS? Neglecting rear-end collisions is a mistake and giving motorists CICAS warnings could increase the number of conflicts leading to rear-end collisions.
• What is the cost-benefit for these systems? The cost-benefit calculation is not clear at this point, but one participant asserted that it should be easier to determine for the infrastructure approach. In seven to ten years it should be easier to predict the status of infrastructure deployment. Another participant asked whether the overall incidence of crashes at signalized intersections justifies the cost of the program. (Although this question was raised, most did not believe this was an impediment to continuing with the CICAS program.)
• Radar and Lidar are expensive technologies and there are associated risks for the OEMs. The vehicle equipment for DSRC will need to be multipurpose in order to provide an adequate ROI.
• How will signal software version upgrades be handled? Upgrades will be easy for integrated systems. VII is integrated; therefore, software upgrades should be easy. Centralized signal systems need to be able to verify that the system is functioning. User expectations combined with infrastructure and vehicle system failures could be a problem when CICAS is operating.
• What is the possibility that vehicle systems can learn driver behavior and set tailored gap acceptance algorithms? One response was that this would be too much liability for OEMs to take on.

CICAS Vision
Mike Schagrin, ITS JPO

The U. S. DOT vision for the CICAS program includes the following:

• Conduct a research, systems integration and development program
• Establish a cooperative system addressing all four problem areas by 2009
• Establish the violation countermeasure first, but develop the gap acceptance countermeasure in parallel
• Employ technology that is close to being production ready
• Work in cooperation with the Vehicle Integration Initiative (VI), especially for field testing
• Push the full range of countermeasures

Attendees believe that there needs to be a mechanism to share ideas and keep track of developments in the infrastructure and vehicle groups. Attendees also believe that there is a need to know what the program is deploying and what the work program will look like. Are we talking about a stopgap measure or are we talking about systems that can be deployed down the road? Attendees also believe that the CICAS program needs to coordinate with VII so it doesn't overlap their efforts. This raised the question: Should VII drive CICAS? There was no consensus on this question; however, the CICAS Program Manager sees the CICAS initiative as an important stand-alone program as well as one that shares resources with the VII initiative.

Questions and comments on the CICAS vision included the following:

• Question: Why do you say that an infrastructure system offers limited benefits? In order to gauge the effectiveness of a vehicle-based system, one needs to factor in the percentage of cars equipped with the technology. Answer: We will only be able to equip a limited number of intersections, so there is a limit to effectiveness for infrastructure-based systems as well. Response: States argue these are high-risk intersections and that this is an interim solution that their politicians can buy.
• Question: Clarify that signal violation is done in parallel with gap acceptance, not before it
• Comment: There has not been enough of an emphasis in understanding the benefits of CICAS; we need to build the case.
• Comment: Cooperative may not necessarily mean DSRC. Answer: DSRC is a key technology, we have a limited time frame and there has been a lot of investment; other technologies cannot support our tough latency requirements.
• Question: What is the evaluation portion of the roadmap? Answer: the evaluation component is represented in the diagram of the CICAS Vision.
• Comment: Violations and gap acceptance have parallel paths; these are two fundamental problems.

Breakout Session #1
The CICAS Vision and Evolutionary Path

The participants were divided into six breakout groups. The groups were asked to use the high-level vision presented by the U. S. DOT and "fill in the blanks." The groups were asked to answer the question: "What does the system look like and when?" They were asked to note the status of development of the system and its components for each of four problem areas. The problem areas were:

• Stop-sign violation
• Signal-violation
• Stop-sign assist
• Left-turn against path/opposite direction (LTAP/OD)

The milestones from the present state in 2005 through intermediate points in 2006 and 2007 to the pre-deployment endpoint of 2009 were timed to coincide with major tests being held for VII. Participants were instructed that by 2009, CICAS should have undergone a field operational test and should be deployment-ready. The components suggested for the discussion were:

• Sensors
• Communications
• Driver interface
• Processing/controllers
• Other

The U.S. DOT vision sets parallel development paths for the two problem area types that CICAS addresses: violations (signal and stop sign) and gap acceptance (stop sign assist and LTAP/OD). Workshop participants agreed with this approach and noted that the two problem area types require different types of countermeasures, and that the gap problem area involves more difficult driver behavior issues.

Participants reached a consensus that the U.S. DOT vision for CICAS has merit. Reasons for this judgment include the immediate availability of U.S. DOT resources for CICAS, the efficacy of an evolutionary approach, and the utility of a total systems approach. Arguments against the vision were that the research program may be overly ambitious, may raise false expectations, and is somewhat artificially confined to intersections.

Driver behavior and human factors, in general, were highlighted in the discussion as a research area that should be emphasized, particularly using naturalistic research designs. Participants cautioned that the promotion of CICAS needs to account for issues such as:

• The unintended consequences of CICAS
• The feasibility of the five-year timeline
• The relationship of CICAS to the VII program
• The role of the infrastructure or vehicle, respectively, in the cooperative system and the appropriateness of an infrastructure- or vehicle-based system for solving each of the problem areas
• The measurement of costs and benefits

Most groups discussed the evolutionary path in terms of the CAMP model presented by automobile manufacturers. Much discussion focused on having the evolution start with an infrastructure-only solution and then follow the path that CAMP presented from Concept 3 to Concept 5[2], but with more functionality resident in the infrastructure than indicated in the CAMP model. Workshop participants judged the 2009 milestone for a post-FOT, deployable CICAS as achievable.

A detailed narrative of the results of the discussions for Breakout Session #1 is given in Appendix A1. A tabular summary of the vision and evolutionary path by milestone and component is in Appendix A2.

Day One Wrap-up Discussion

• The breakout sessions were valuable.
• The follow-up task is to take the CAMP model, add infrastructure to it and discuss the result in the next workshop
• One example of the functionality that infrastructure-based systems add (i.e., vision-based systems) is pedestrian detection.

Day Two – 12/09/2004

CAMP Review

CAMP reviewed the seven models (also called concepts or scenarios) in more detail. CAMP said that the major technology gap areas were the following:

• Positioning
• Sensing
• Computing
• Driver Interface
• Communication

The discussion revolved around understanding the relationship of the CAMP concepts to each other, to the problem areas, and to infrastructure-based CICAS components and solutions. CAMP has focused on vehicle-based systems and has not addressed the infrastructure side in detail. Participants asked: Why not look at the CAMP evolutionary path and overlay the infrastructure evolutionary path?

• Relationships Among the Scenarios and the Problem Areas: Were the seven scenarios scoped out before the problem areas were addressed? CAMP replied that all problem areas were taken into account when the scenarios were created. Concept 1 covers all problem areas.
• Concept 2 Light: In this scenario in-vehicle signing to vehicles within some radius of the intersection would not require access to vehicle services. It was suggested that an aftermarket device could potentially deliver the message to the driver – e.g., a PDA connected by Bluetooth to the vehicle.
• Definitions of Warning, Driver Alert, and Advisory: There was discussion on the nature of the information delivered to the driver so that the driver can make the right decision. The DII will tell the driver that it is not safe to go. What happens when the system is not working? Terms are defined in the MUTCD.
• Concepts 2, 3, and 4: Concept 3 is a lower cost measure, which only requires an RSU DGPS, DSRC and some processing; there is some concern regarding channel capacity. Concept 3 does not have the capacity problem. Concepts 2 and 3 are not mutually exclusive. Concept 3 assumes no sensors in the infrastructure; the sensing would be done through RSUs. Concept 3 does not take care of the "stop-sign-assist" problem. Stop-sign-assist is really a Concept 4 and 5 problem. Concept 3 is actually a lower cost equipment measure involving the RSUs providing DGPS, DSRC, and processing power.
• Concept numbering: The numbering of the concepts implies a unidirectional flow of development, but there is a need to define parallel lines of development. One suggestion was to reverse Concepts 3 and 4.
• Relationship Between CAMP and Infrastructure-based Systems: The non-CAMP researchers expressed a need to understand what the CAMP concepts are from the infrastructure point of view. PATH suggested that from the intersection point of view the vehicle is just another sensor and vice versa. VTTI asked: In a mature system what does CAMP need from the infrastructure, i.e., in the vehicle-to-vehicle scenario what is needed from the infrastructure? CAMP replied that they need the geometry of the intersection (precise navigational map, including stop bar locations, and characterization of the individual lanes, such as the left turn lane). A state map (which will provide vehicle ID, lane-mapped position, distance from intersection and speed) would also need to include vehicle type – which can only be precisely determined by vehicle-based systems. The V2V message set Society of Automotive Engineers (SAE) standard is currently in draft. Schagrin stated that we know what we need, but we don't know the precision. Another comment to CAMP was that stop sign assistance requires a heavy investment in infrastructure sensors.
• Weather Information: There are not currently very many road weather sensors, so we don't expect to get needed weather information from Road Weather Information Systems (RWIS).
• Maintenance Requirements: A comment from the States was that we need to take into account during system design the maintenance load for state and local highway departments that would be increased by the additional capabilities in the RSU. This is an integration issue.
• Definition of Traffic Control Device: The University of Minnesota noted that we should enlarge definition of traffic control device to be an intersection control system.
• Liability and Other Policy Issues: The policy committee of AASHTO and the automobile manufacturers are looking at liability issues. These policy issues may be outside the scope of this CICAS group. There is overlap with issues that VII is addressing, e.g., liability and RSU rollout.
• Research questions to be answered:
• What are human factors issues? There is a conflict if the warning comes from both the infrastructure and the vehicle; there may be unintended consequences of the warning; naturalistic data are needed.
• We are assuming DSRC will be effective, but how will it perform in urban areas, around corners, etc.?
• What will drive DSRC OBU deployments? We expect non-safety applications to accelerate deployments. One project, recently ended, implemented standard SAE messages. The system was never prototyped, but it is now ready. We need sensors on the SAE bus. There is storage on the RSU and it is needed also on the vehicle.
• What type and quality of information do drivers need to make safe intersection decisions?
• How will our systems account for pedestrians (50 percent of crashes involving pedestrians occurred at intersections) and bicycles (70 percent of crashes involving bicyclists occurred at intersections)?
• We need models of safe gaps and driver intent.
• What are the consequences of system failure?
• What are the requirements for signal controller interfaces?
• We need system-level requirements.
• How can we achieve accurate lane-level positioning?
• We need to refine and validate the algorithms.
• What level of control is needed for various situations? Levels of capability include:
• Advise
• Warn
• Control
• Enforce

Breakout Session #2
Research Needs and Priorities

The six Breakout Groups discussed the research needs that must be met in order to achieve the U. S. DOT vision for CICAS and beyond that, a mature operational system. Participants were asked to evaluate the level of priority (high, medium, or low), start dates, and dependencies of identified research needs. Workshop participants determined that the highest priority needs were to:

• Characterize normal and non-normal driver maneuvers at intersections in varied, real-world test conditions.
• Determine alert timing parameters
• Determine effectiveness of various interface approaches and their interaction with varied timing approaches
• Develop design guidelines and performance specifications for DII
• Measure effectiveness of DII
• Determine preferred modality (combination of auditory, visual, and haptic alerts) of DVI
• Refine message content and timing
• Add CICAS capabilities or add CICAS interfaces to legacy and advanced traffic controllers
• Determine requirements for accuracy of vehicle position
• Determine requirements for accuracy of vehicle speed, acceleration, braking status, heading angle by in-vehicle sensors, by positioning systems
• Specify infrastructure-vehicle communications
• Design CICAS protocols to be run over DSRC; design DSRC MAC layer protocols for high density traffic conditions
• Improve methods to fuse data from multiple sensors on vehicle position
• Develop understanding of false alarm rate, miss rate, driver acceptance of computations

A summary of the high, medium and low research priorities that emerged from Breakout Session #2 is in Appendix B1. A tabular summary of the discussion is given in Appendix B2.

Brainstorming Session
Environmental Scan

As a final discussion activity, workshop participants discussed the environmental factors that will affect CICAS development, both positively and negatively. Factors were grouped into four cells depending on whether they derived from the public or private sectors, and whether the issues were systems/technical or environmental/institutional. Figure 1 shows the results of the environmental scan.

Next steps:

• Next workshop: February 17–18, 2005
• The Infrastructure Consortium will craft an infrastructure evolutionary path, reach consensus, and work with CAMP to consolidate into a unified, comprehensive evolutionary path. Gary Allen and VDOT have agreed to lead this effort.
• The vehicle/infrastructure combined evolutionary path will be purely conceptual at this time.
• The ITS JPO requires a first draft high-level concept of operations and functional requirements, as well as system architecture.

Footnotes

• Table 2 was abstracted from the materials prepared by CAMP. Back
• Concept 3 is where the infrastructure provides data to the vehicle, which makes the threat assessment and delivers a warning through a DVI. In Concept 5 both the vehicle and the infrastructure provide data to the vehicle's threat assessment processing and delivers the warning through both a DVI and DII. Back