A bi-level hybrid algorithm for solving multi-target inspection path planning problem of mobile robot in complex radioactive indoor environment

[Display omitted] Multi-target inspection path planning (MTIPP) of mobile robot (MR) represents a significant area of research in the context of environmental monitoring and routine inspection of nuclear power plants (NPPs). Given the challenges posed by complex radioactive indoor environments, characterized by the presence of numerous radioactive sources and dense obstacles, a bi-level multi-objective programming framework is proposed to model the MTIPP problem. To navigate this model effectively, a novel bi-level hybrid algorithm named ACO-GA-A* that integrates improved ant colony optimization (IACO), genetic algorithm (GA) with modified A* algorithm is developed. In the upper level, ACO with a GA-based non-uniform initial pheromone distribution, an adaptive heuristic function and an elite strategy for pheromone update is employed to determine the optimal traversal sequence of inspection targets. In the lower level, a modified A* algorithm, which considers multiple constraints including path length, risk degree and energy consumption, is utilized to plan pairwise paths between targets, thereby generating cost graphs. Comparative simulation experiments are conducted in various complexity radioactive scenarios. The results indicate that the modified A* can plan pairwise paths with lower total costs in shorter time compared to traditional A*, ACO, and GA. Furthermore, the ACO-GA-A* demonstrates better sensitivity, reliability, and convergence characteristics compared to some other bi-level hybrid algorithms. Subsequent real-world experimentation corroborates the effectiveness and feasibility of both the bi-level programming framework for MTIPP and the proposed ACO-GA-A* algorithm.