Cooperative vehicle sensing and obstacle avoidance for intelligent driving based on bayesian frameworks

Vehicular Adhoc Networks (VANET) based vehicle sensing and obstacle avoidance is of importance and widely addressed in intelligent driving. Due to the difficulties in the data fusion from various types of observations from different vehicles, a dynamic non-parametric belief propagation (DNBP) method based on the Bayesian framework for target detection and localization is presented. Furthermore, the target detection performance can be jointly improved by adopting observations from multiple vehicles, based on the presented frameworks. The presented method is validated through simulations. The performance advantages achieved from joint detection from multiple vehicles are also evaluated.