By integrating DOT crash data with years of driving behavior data, we created a clear map of the riskiest areas for VRUs across Nevada.For the VRU assessment, Michelin Safer Roads looked at 6.4 billion trip points, 13.4 million acceleration events, 33.6 million braking events, and 26,947 miles of road. Using guidelines provided by the U.S. Department of Transportation Federal Highway Administration, we analyzed data including location, roadway functional classification, design speed, speed limit, and time of day.
To offer an additional layer of insight, a cluster analysis was performed on the data based on similarity with AI and machine learning and used a potential risk model to identify potential VRU risks.
Michelin Safer Roads created VRU Risk Maps ranking the severity of each county in Nevada. Nevada DOT was provided insights on the highest-risk areas for VRUs within each county.
Focusing on 10 areas across Clark, Washoe, and Carson City Counties, Michelin Safer Roads indicated which specific areas were riskiest for VRUs so Nevada could begin to deploy safety countermeasures.
For example, past historical crash data showed a high number of crashes on multiple roads in Clark County. Our insights indicated the high risk for VRUs were on busy sidewalks that had very high pedestrian volume, but did not have buffers or safety strips protecting them from traffic. The proximity of retail stores and restaurants also raised the likelihood of pedestrians crossing at mid-block or during red lights.
In Carson City County, large sections of the road network were found to have higher VRU risk, even in intersections with pedestrian crossings and dedicated bike lanes. Historical data indicated VRUs had been involved in crashes in the area. Speeding was also found to be a contributing factor for risk.
With this thorough and objective analysis, Nevada DOT is better positioned to recommend mitigations in each area. They now also have the data and insights they need to justify and explain safety investments and future funding needs.
