Cooperative Intersection Collision
Avoidance Systems (CICAS)

CICAS Workshop

December 9–10, 2004
Appendix B1
Summary of Breakout Session #2: Research Needs

High Priority Research Needs

• Characterize normal and non-normal driver maneuvers at intersections in varied, real-world test conditions.

• Need to determine what data is available for mining and what data needs to be collected (e.g., the 100 car study)
• Need for development of data collection systems and amplified naturalistic driving conditions through a set of intersections
• Need to evaluate and develop collection techniques (e.g., observational, infrastructure and vehicle sensing)
• Determine gap selection
• This will be different for various intersections, but will be less of an issue for simple scenarios
• Some work being done by CAMP and will be public soon
• Must start selecting intersections now in order to get baseline data for analysis in 05/06
• Can begin immediately especially benchmarking of available technologies for flexible implementation
• Dependent on delivery surrogate safety measures a system for collecting the SSM data
• This is needed to create the state map and should be started now
• Dependent on the sensor suite used
• Dependent on legacy system upgrades, vehicle data comm., control strategies

• Determine alert timing parameters

• Need to define what are we warning the drivers about
• Dependent on characteristics of normal and non-normal driver maneuvers at intersections in varied, real-world test conditions
• Need a better handle on "acceptable safe vs. unsafe" gaps
• Current algorithms need to be refined based on how the driver is behaving/driver characteristics data

• Determine effectiveness of various interface approaches and their interaction with varied timing approaches

• Coordinate and build on current work including, MN & CA efforts in FY 05
• Use information from previous FOT
• DVI work being done by CAMP, VTTI, others
• Need to work on the issue regarding the vehicle not sending information to the infrastructure for it to make a decision

• Develop design guidelines and performance specifications for DII

• Need an operational test system first because effectiveness cannot be measured until prototypes are selected
• Tests should be performed on a small number of the intersections pre-FOT
• This should only be done in conjunction with DVI

• Measure effectiveness of DII

• Need an operational test system first because effectiveness cannot be measured until prototypes are selected
• Tests should be performed on a small number of the intersections pre-FOT
• This should only be done in conjunction with DVI
• Work should start now and will require iterations
• First test effectiveness in driving simulator then test in "live" intersection
• Dependent on current research results

• Determine preferred modality of DVI (combination of auditory, visual, and haptic alerts)

• There is already good knowledge of this and existing research findings should be leveraged, as well as ongoing work
• Evaluate after the test vehicle is built
• This should only be done in conjunction with DVI
• Some work already underway, but more research needs to be done
• Not sure this is applicable to CICAS or is this an OEM issue

• Refine message content and timing

• Need to continue current efforts and refine concepts
• Cannot refine until first message sets defined
• Each application will require different message sets with overlap
• This is an OEM issue

• Add CICAS capabilities or add CICAS interfaces to legacy and advanced traffic controllers

• Requires specifications on timings
• Need to provide this critical data NOW!

• Determine requirements for accuracy of vehicle position

• Need to develop system architecture and requirements for both infrastructure and vehicle
• Define levels of accuracy absolute/lane level/relative
• VTTI has developed some specifications, need to begin providing this for SCP
• This requires a new detector system

• Determine requirements for accuracy of vehicle speed, acceleration, braking status, heading angle by in-vehicle sensors, by positioning systems

• VTTI has developed some specifications, need to begin providing this for SCP
• Important for infrastructure only models, because this is an infrastructure sensor issue
• Need requirements for measuring accuracy of all sensors and need concept of operations
• Vehicle position and kinematics dependent on naturalistic data, other data, prototypes, functional requirements
• Is iteration of alert algorithms what follows? Need to answer, "What sensors can get you what you need?"

• Specify infrastructure-vehicle communications

• Need to understand budget for deployment
• Need to answer, "What role does DSRC play here?" Need to start discussions about DSRC Message Frequency, testing, etc.
• OK to assume DSRC is short-term
• Specify communication parameters, but not specific technology
• SAE message committee final specification follows testing
• CICAS protocols should start now
• Need to standardize the message set

• Design CICAS protocols to be run over DSRC; design DSRC MAC layer protocols for high density traffic conditions

• Let's see if it works first
• Given deployment schedule will be 802.11 xxx anyway
• Much of this work is being done in VII
• DSRC is NOT proven and this will be an issue
• Communication issue, VSC2 will do this
• Systems architecture and requirements needed
• Message set is main issue

• Improve methods to fuse data from multiple sensors on vehicle position

• Operational systems must precede this event and must be done prior to determining the worthiness of the project
• For mixed vehicular/infrastructure this is key, but need to solve same detection problems
• Challenging problem…must work with sparse and dense data
• Vehicle data depends on DSRC deployment
• Functional allocation dependent on data from vehicle
• Cost effectiveness to put inside a vehicle is important

• Develop understanding of false alarm rate, miss rate, driver acceptance of computations

• Should occur after FOT
• Prototype subsystem needed now and available test bed
• Second phase issue: need to evaluate early test results
• Dependent on driver reaction to technology reliability
• Hard to classify "hit" or "miss" when no event occurs

Medium Priority Research Needs

• Investigate the effects of driver distraction and inattention

• Dependent on application
• Dependent on current data collection efforts (e.g., the 100 car study)
• Long term data collection process needed

• Investigate long-term driver behavioral adaptation, either positive or negative

• Need an operational system first
• Evaluate prototype before we can start the adaptation issue
• Needs to be examined long term, greater than 5 year period
• This data will be provided by the FOT
• Methodology needed for FOT design
• The DII needs to be optimized
• Depends on current research results and mining of national databases
• This may provide a baseline investigation on what worked
• Key point is none appear promising, not sure that existing data can be mined to extract needed information, I don't think it will give any significant information
• This is a cheap and easy way to increase the value of the databases, biased towards intersection behaviors
• May help immediately with needs identification for draft functional requirements
• Need to identify data requirements and data collection methods
• Need the development of a more advanced infrastructure data collection system and to run a naturalistic driver study through a set of monitored intersections
• Look at 100 car study 43,000 hours, 100,000's of intersections (of all kinds) crossing accessible in short term

• Detection of pedestrians and bicyclists in the intersection

• Research needs to be done for detection
• Need a long lead time and this is technically very challenging
• See UMN work
• Applies only to GAP acceptance
• Is this an ITS Tier 2 issue

• Determine dynamic power and data rate control strategies

• DSRC and VII design issue
• Dependent on high DSRC penetration
• Being worked on, VII issue with ASME
• Need prototype DSRC radio first

• Define CICAS message sets and dictionaries for submission to standards developing organizations

• Need to get concepts refined for detailed design
• Message sets define further work

• Determine whether the performance characteristics of various systems architecture meet specified requirements

• Need to follow a Systems Engineering process
• This is not well defined and depends on concepts refinement then analysis of architecture and specifications

• Measure CICAS safety benefits in comparison to other traffic control devices

• No way to accurately model without well designed FOT
• Cannot do until CICAS performance is quantified (otherwise just a guess)
• Need this before the FOT and add as baseline condition
• Measure CICAS benefits in comparison to unsignalized intersections

• Determine costs and benefits of alternative CICAS approaches

• Need available data, especially from car companies and OEM's, ROI, User Demand
• Cannot do until optimal set is deployed and costs determined

Low Priority Research Needs

• Refine economic impact models of intersection collisions

• Need for cost benefits analysis
• No way to model without well designed FOT
• Need this before the FOT