A Multi-robot Path Planning Approach For Nuclear Decommissioning Scenarios

Nuclear decommissioning is a crucial part of the complete life cycle of the nuclear industry in China,and the entire implementation process is in a radioactively contaminated nuclear radiation environment.In order to speed up the entire nuclear decommissioning cycle and reduce the threat to human health from manual work in special environments,it is practical and feasible to replace manual work in nuclear contaminated areas by robots,which is also an inevitable trend for the industry.Multiple robots have the advantages of high efficiency and robustness compared with single robots,and the application scenarios of multiple robots are more extensive and have certain generality.The path planning method is the key technology to complete the deployment of multi-robot function,and also belongs to the focus of multirobot application research.The multi-robot path planning method mainly consists of three parts: multi-robot task assignment,global path planning and local path planning.The specific research contents are as follows.(1)Multi-robot path planning system solution design for nuclear decommissioning scenario.According to the nuclear decommissioning context in the subject,considering that the application environment of the robot contains both physical environment with obstacles and nuclear radiation environment contaminated by radioactivity,it is determined that the environment map model is established by the raster method and the nuclear radiation data is simulated with the assistance of Geant4.The vehicle path model with capacity constraints in the nuclear decommissioning scenario was established for the task requirements,and the multi-robot control method and path distance estimation method were determined,which laid the foundation for the subsequent multi-robot task assignment and path planning.(2)Multi-robot task assignment method in nuclear radiation environment.For the problem of high overall nuclear radiation dose in multi-robot operations in nuclear environment,an improved genetic algorithm is proposed as a multi-robot task assignment method,an adaptation function is designed with the overall nuclear radiation dose magnitude as the evaluation criterion,a more efficient crossover operator is improved from the task assignment model,and a local search operator using a large-scale neighborhood search algorithm is introduced.Experiments show that the improved genetic algorithm has significant advantages over other heuristics in terms of convergence accuracy,speed and stability.(3)Multi-robot path planning method in nuclear radiation environment.For the problem that the traditional A* algorithm searches global paths existed against the wall and near the high radiation region,the evaluation function of A* algorithm in nuclear radiation environment at the cost of low radiation dose of paths is redesigned,and the artificial potential field method and dynamic coefficients influenced by radiation sources are introduced.For the path conflict problem existing in multi-robot local path planning,the dynamic window method containing priority waiting strategy is used to solve it.Experiments show that the proposed algorithm can efficiently plan a cumulative low-radiation dose global path without risk points as a way to improve the service life of multiple robots,and the number of search nodes is reduced by50.04% on average compared to the comparison of the improved A* algorithm,and the number of risk nodes are all reduced to 0.The improved dynamic window method can effectively solve the path conflict problem.Finally,according to the proposed multi-robot path planning method in nuclear decommissioning scenario,physical experiments were conducted in the laboratory site using three Songling robots Scoutmini.The results show that the multi-robot task assignment is efficient,the path planning has low accumulated nuclear radiation dose,and the path conflict can be resolved by real-time obstacle avoidance.The proposed multi-robot path planning method is of reference value for robotic applications in nuclear radiation environments.