A Two-level Path Planning Method for On-road Autonomous Driving

We present a real-time path planning method for on-road autonomous driving. This method utilizes a two-level hierarchy consisting of on-road behavior planning and online path generating to achieve feasible paths for autonomous vehicle in dynamic on-road environment which is characterized by lots of structured information and flat terrain. An on-road behavior planning technique is used to transform the uncertain path planning problem into several determined sub-ones. Then, the online path planning algorithm is applied to solve these sub-problems. This method contains three major components: anytime behavioral replanning, online path planning and mitigation strategy for time lag due to an agent system with asynchronous multi-sensor. We evaluate the proposed method on the KUAFU-1, an autonomous vehicle prototype modified by ourselves. Experimental results show that the proposed method can successfully provide real-time feasible paths in various on-road traffic sceneries.