Leveraging Neural Networks to Guide Path Planning: Improving Dataset Generation and Planning Efficiency

Recent advances in Rapidly-exploring random tree: A review

“…Baldoni et al [ 175 ] utilize neural networks to guide the generation of path planning with the expectation of improving the generation of datasets and increasing the efficiency of path planning. The authors conclude that the path planning task for narrow passages and maze-like maps is very similar to the image segmentation task, so they choose U-net, which is widely used in the field of image segmentation, to train the path planning task, and the results show that the neural network-guided RRT significantly outperforms the traditional RRT in terms of path planning.…”

“…(5) For the similar specific problems in the field of artificial intelligence and the field of path planning, the methods involved in the field of artificial intelligence may also be directly used in the field of path planning to achieve good results, for example in Ref. [ 175 ], the authors fuse U-net into RRT due to the similarity between path planning for maze-like maps and high-precision image segmentation. (6) An interesting work is to generate unpredictable paths to ensure privacy, a time when supervised adversarial neural networks show their strengths [ 176 ].…”