1
2	V2V Safety Framework
3	Connected Vehicle Environment Full deployment – all vehicles capable of communicating with each other Model Deployment – subset of all vehicles capable of communicating with each other
4	Scoping the Model Deployment Question: What should be the scope of the Model Deployment to gather enough data? Conducted analysis using prior field test results 3 Forward Collision Warning alerts during treatment period
5	Results of Scoping Analysis Requires careful selection of test area Requires careful selection of test participants as well to ensure interactions
6	Experimental Design – Recruitment Approaches for Recruiting Participants UoM Medical Center Ann Arbor city school system Variety of Interactions Following Adjacent Crossing
7	Simulating the Experimental Design Ann Arbor Trip Tables Provided by Washtenaw Area Transportation Study TRANSIMS Model Ran model for 24 hour period Output included second-by-second positioning of vehicles
8	Post Processing to Obtain Interactions Forward collision Same direction, same lane Lane change Same direction, adjacent lanes Intersection assist Crossing paths
9	Interactions by Time of Day 64 light vehicles equipped with safety applications 2,500 Vehicle Awareness Devices deployed Hourly estimates of interactions in a typical weekday
10	Estimates of Daily V2V Interactions Assumptions for V2V interactions 2,500 Vehicle Awareness Devices are deployed Vehicle speeds > 25 mph Vehicles are within 30 meters of each other
11	Total Safety Alerts Estimated How do the interactions relate to safety alerts generated by the applications? Developed estimates of the relationship between interactions and safety alerts. Estimated that over 6 months, each driver would experience ~3 alerts / safety application
12	Interaction Results from Month 1 Devices Deployed 64 Integrated light vehicles are deployed 738 Vehicle Awareness Devices are deployed
13	Conclusion Observed results from the field are comparable with the simulated estimates Simulation Model estimated that the Model Deployment will generate sufficient data for evaluation