摘要：AbstractSampling-based motion planning methods are widely adopted in autonomous driving. Typically, sampling can be decoupled into two layers: a path sampling layer and a speed profile sampling layer. For the path sampling layer, traditional methods tend to sample with a uniform distribution over the whole feasible space, which might cause either computational inefficiency or poor performance if the sampling resolution is not set appropriately. To solve this problem, we propose an adaptive path sampling approach that samples from a time-varying distribution depending on the dynamic environment and potential costs of the ego vehicle. Such sampling strategy is then combined with a non-conservatively defensive strategy in the speed sampling layer to generate a set of safe but not overcautious trajectories. The proposed motion planning framework is tested both in simulation and a real autonomous vehicle in a roundabout scenario. The results demonstrate that it can efficiently generate non-conservative but defensive trajectories to safely drive the vehicles in dynamic environments full of uncertainties.
