Preceding Vehicle State Prediction Using V2V Safety Messages

In this paper, we present a framework to predict the future state of the immediate preceding vehicle of an ego vehicle in response to other lead vehicles on the road. We propose a three-phase method to fuse measurements from onboard sensors with wireless data received from lead vehicle(s) when connectivity penetration rate is low. That is, we estimate the link length between the preceding and lead, add virtual vehicles to the platoon to simulate the motion of unconnected vehicle, and predict the future state of the preceding vehicle up to 5 sec in the future. Using experimental dataset, we show that with only one connected lead vehicle, the error in predicting the velocity and position of the preceding vehicle can be improved by 54.69% and 39.65% while reducing the standard deviation with 54.13% and 39.12%, respectively, compared to the case with no connectivity.