Multi-robot coverage path planning for dimensional inspection of large free-form surfaces based on hierarchical optimization

The multi-robot coverage path planning problem (MRCPP), incorporating with the sub-problems of viewpoint sampling, viewpoint task allocation and robotic sequential planning generally, is a highly coupled problem with complex engineering constraints and is difficult to achieve the optimal result. Most coverage path planning studies in the field of industrial inspection focus on viewpoint sampling and sequential path planning for the individual robot. At the same time, the measurement uncertainty requirements of the inspected surface features for the candidate viewpoints are rarely considered in the process of CPP. To address this problem, a systematic MRCPP framework for free-form surface inspection is developed considering the measurement uncertainty requirements and full coverage of the surfaces to ensure the inspection efficiency and the scanning accuracy of to-be-inspect features. Specifically, a stepwise method for multi-robot optical coverage path planning is proposed. Firstly, a redundant viewpoint set is generated based on the geometric model of the product; additionally, considering the constraints of measurement uncertainty and full coverage constraint, an enhanced Random Inspection Tree Algorithm (RITA) is used to obtain the optimal viewpoint set. Secondly, an improved group Ant Colony optimization (IACO) algorithm is proposed to realize multi-robot viewpoint allocation and sequential planning on the optimal viewpoint set. Finally, in order to evaluate the effectiveness of the proposed method, a 4-robot optical inspection case of an auto body was used. Compared to the benchmark method, the number of viewpoints is reduced by 56.38%, and the bottleneck time in the station is reduced by 7.651%.