Multi-Robot Path Planning Based on Lens Imaging Reverse Learning Harris Hawk Algorithm in Dynamic Environment

To overcome the difficulties of avoiding obstacles in real time and being vulnerable to local optima in multi-robot path planning (MRPP) in unfamiliar locations, a lens imaging reverse-based learning Harris Hawks optimization algorithm (LRHHO) is proposed. Initially, the Latin hypercube sampling method is employed for population initialization to enhance the diversity and uniformity of the population. Subsequently, during the local exploitation phase, the lens imaging reverse-based learning strategy is introduced to refine individual position updating mechanisms. It is complemented by an adaptive roulette wheel mechanism designed to select between current solutions and their reverse-based counterparts. Finally, a restart mechanism is implemented for inferior individuals to improve the overall evolutionary efficacy of the population. Comprehensive evaluations on benchmark test functions demonstrate the superior optimization performance of LRHHO compared to existing algorithms. A real-time MRPP model is constructed utilizing relative positioning methods, where LRHHO optimizes robots' velocity and angular parameters to determine subsequent positions. The system integrates three coordinated obstacle avoidance mechanisms: static obstacle avoidance, dynamic obstacle avoidance, and inter-robot coordination, enabling rapid responses to randomly moving and unforeseen obstacles. Simulation experiments conducted in two scenarios with varying complexity levels reveal that the proposed method achieves intelligent real-time obstacle avoidance while coordinating concurrent multi-robot operations. Comparative results indicate that LRHHO generates higher-quality paths with enhanced efficiency relative to alternative algorithms.