Fast Trajectory Planning for Off-Road Autonomous Driving with a Spatiotemporal Tunnel and Numerical Optimal Control Approach
Bai Li, Youmin M. Zhang · 2019
This paper proposes a unified and fast trajectory planning method for an autonomous vehicle with avoidance of static obstacles in the off-road scenarios. Nominally, an optimal control problem should be used to cover generic off-road planning cases. However, the scale of the formulated optimal control problem is intractable, because the ego vehicle has to avoid collision with every obstacle at every moment. To address this issue, we aim to discard the redundant collision-avoidance constraints, because the vehicle may not really have chances to collide with all of the obstacles at every moment. Concretely, with the help of a reference trajectory derived by hybrid* algorithm, we develop a spatiotemporal tunnel-based strategy such that the ego vehicle is restricted to move within a sequence of tunnels during different periods of the entire dynamic process. Through this, the collision-avoidance constraints are replaced by the within-tunnel constraints, thereby making the scale of the optimal control problem completely irrelevant to the complexity of the environment.