Heterogeneous Multi-Agent Coverage Control through Adaptive Weighting and Corrective Potential Function

Abstract This paper presents a novel adaptive coverage control strategy for multi-agent systems with dynamic weight learning and obstacle avoidance. The approach combines power diagrams with adaptive weight evolution and an improved artificial potential field (APF) for intelligent agent navigation. Each agent dynamically adjusts its power diagram weight based on real-time sensing quality, local environment complexity, and resource availability, while being attracted to the centroid of its adaptively-sized power cell. The system ensures robust coverage and collision avoidance through distributed learning mechanisms that adapt to changing environments and agent capabilities. Theoretical analysis provides formal convergence guarantees for the adaptive weight dynamics, and the improved APF eliminates local minima that plague traditional potential field methods. Comprehensive validation through extensive simulations demonstrates superior performance in challenging, large-scale scenarios with high obstacle density.