1: \begin{abstract}

2: Balancing the trade-off between safety and efficiency is of significant importance for path planning under uncertainty.

3: Many risk-aware path planners have been developed to explicitly limit the probability of collision to an acceptable bound in uncertain environments.

4: However, convex obstacles or Gaussian uncertainties are usually assumed to make the problem tractable in the existing method.

5: These assumptions limit the generalization and application of path planners in real-world implementations.

6: In this article, we propose to apply deep learning methods to the sampling-based planner, developing a novel risk bounded near-optimal path planning algorithm named neural risk-aware RRT (NR-RRT).

7: Specifically, a deterministic risk contours map is maintained by perceiving the probabilistic nonconvex obstacles, and a neural network sampler is proposed to predict the next most-promising safe state.

8: Furthermore, the recursive divide-and-conquer planning and bidirectional search strategies are used to accelerate the convergence to a near-optimal solution with guaranteed bounded risk.

9: Worst-case theoretical guarantees can also be proven owing to a standby safety guaranteed planner utilizing a uniform sampling distribution.

10: Simulation experiments demonstrate that the proposed algorithm outperforms the state-of-the-art remarkably for finding risk bounded low-cost paths in seen and unseen environments with uncertainty and nonconvex constraints.

11: \end{abstract}

12: