Efficient and High-Quality Path Planning Guided by Map Information for Mobile Robots

Path planning of mobile robots plays an important role in improving work efficiency in the manufacturing industry. On one hand, the sampling-based path planning methods commonly used in mobile robots require extensive and time-consuming iterative searches to find a relatively optimal path due to their stochastic nature. On the other hand, the map information obtained during planning is solely utilized for collision detection without being effectively leveraged. This article proposes a map-informed rapidly random-exploring tree (MIRRT) algorithm that combines uniform sampling, rapidly crossing obstacles (RCO) and progressive optimization. It can utilize map information to quickly guide the search tree. MIRRT involves a preprocessing step that requires segmenting concave obstacles into convex obstacles to prevent path loss. During the growth of the search tree, the RCO method guided by map information can rapidly discover high-quality ground path nodes and efficiently navigate obstacles when they block the path. Finally, simulations and physical experiments were performed to prove that MIRRT significantly outperforms existing state-of-the-art sampling-based planners designed specifically for mobile robots.