Section: New Results

Perception for Cooperative Driving

Participants : Raoul de Charette, Carlos Flores, Francisco Navas, Fawzi Nashashibi.

In cooperation with the control/planning group the computer vision group has worked on practical and applied research for 2D processing of LiDAR sensors in the context of cooperative driving. Practical failures cases were addressed such as the case of occluded vulnerables. In a dense urban environment were buildings may occlude pedestrian we proposed for example a perception system fusing both LiDAR and communication data retrieved from pedestrian communication streaming there GPS position. This allows us to detect and predict possible collisions of car and pedestrians. Experiments were conducted in the site of Rocquencourt and the results lead to a submit journal publication in cooperation with Vicente Milanés from Renault. Another practical case of failures in cooperative driving occurs are the cut-in or cut-out cars in platoon scenarios. When cars travel in a platoon, a car leaving or entering may disrupt the whole platoon. In collaboration with the control group, the detection and prediction of such behavior was addressed using 2D LiDAR data and tested on Cycabs. A journal was submitted in cooperation with Vicente Milanés from Renault.