Path planning of indoor unmanned vehicle based on adaptive A* algorithm

Autonomous vehicle technology, underpinned by advanced artificial intelligence(AI), has become increasingly critical in modern engineering systems. To address the limitations of conventional A* algorithms specifically their computational inefficiency and limited robustness in indoor unmanned vehicle path planning this paper presents an enhanced adaptive A* algorithm driven by AI methodologies. The proposed approach incorporates four key innovations: heuristic function optimization through the integration of an obstacle risk coefficient and turning angle cost to improve path safety and smoothness; dynamic weight adjustment via an exponential decay function to balance search efficiency and solution quality; parent node optimization using recursive cost computation to minimize redundant calculations; and child node pruning to eliminate hazardous neighboring nodes, thereby preventing paths from passing dangerously close to obstacles. Through theoretical convergence analysis, Matrix Laboratory(MATLAB) simulations, and Robot Operating System(ROS) based physical experiments, the algorithm demonstrates significant improvements over traditional methods, reducing turning points by approximately 15%, decreasing planning time by nearly 50%, and substantially optimizing path length. Although the generated paths may not achieve optimality in every specific metric, the method maintains robust performance across environments with obstacle densities ranging from 10% to 35%, effectively enhancing operational efficiency, safety, and environmental adaptability for unmanned vehicle systems. These AI-powered enhancements not only improve the algorithm’s performance in simulation but also demonstrate practical applicability in real world scenarios such as logistics sorting, warehouse patrols, and indoor autonomous navigation, offering a reliable and efficient path planning solution for intelligent unmanned systems.