Search

​

Automated-Vehicle (AV) technologies are expected to improve road safety by detecting the surrounding environment with the equipped sensors (e.g., camera, radar) and making necessary driving decisions. It is worth noting that AVs may not be able to detect all the unsafe conditions due to the limitations of the detection range and/or detection accuracy. The on-board sensors might not be able to detect or classify a target object if it is far away from the vehicle or occluded by other road objects, which might lead to a crash. Pedestrian crashes are more likely to fatal or severe injury crashes. Since 2009, the number of pedestrian fatalities has been rising, reaching over 6,200 in 2019 with an increase of 51%. This project aims to explore the effects of cooperative perception on avoiding pedestrian crashes for connected and automated vehicles. A cooperative perception system will be developed based on a co-simulation platform of virtual simulation (e.g., CARLA) and microsimulation (e.g., SUMO). The cooperative perception is programmed in the simulation platform. The safety benefits of cooperative perception will be evaluated and quantified under different conditions. Besides, to mitigate the data transmission load, data analysis will be conducted to develop statistical models or machine learning algorithms, which could help determine when the cooperative perception is needed and how the perception data from roadside sensors should be fused.