Robot Navigation in Obstacle Environments Based on Spatio-Temporal Analysis

In this paper, we propose a robot navigation method based on proto-path analysis and position change reward, which captures multiple kinds of dynamic interactions in the environment through spatio-temporal graph modelling and meta-path analysis to optimize the navigation decision. The position change reward guides the robot to explore in unknown regions, effectively improving navigation efficiency and avoiding unnecessary paths. Experimental results show that our method can achieve better performance in complex scenarios.