Active safe motion planning for intelligent vehicles in dynamic environments

Abstract

Motion planning is an essential component in intelligent vehicle study. Rapidly-exploring Random Tree(RRT) and its variants are popular algorithms that have been successfully applied in solving motion planning problems. However, obtaining an optimal trajectory while concerning driving safety in dynamic environments is a difficult problem. In this study, we present an active safe RRT(AS-RRT) motion planning algorithm that enable the intelligent vehicle to avoid collision risks and find an efficient path in the dynamic environment. The algorithm firstly reconstructs a potential field-based configuration space for static obstacles and moving vehicles, which defines the risk regions. Then, it develops an RRT tree through samples in the space with considerations of nonholonomic constraints of the vehicles. A comprehensive cost function is used for the priority sequence mechanism to get an initial trajectory. After that, the trajectory is asymptotically optimized gradually by decreasing the cost iteratively. Simulation results demonstrated that the proposed algorithm improved the vehicles’ motion planning safety performance in dynamic environments.