To address the issues of unbalanced residual energy caused by heterogeneous initial robot states and dynamic environmental disturbances, this paper proposes a dynamic task allocation and rescheduling strategy considering energy balance. A Multiple Traveling Salesman Problem (MTSP) mathematical model that incorporates energy constraints and load balancing is established. Furthermore, an Improved Genetic Algorithm (IGA) based on K-Means initialization and adaptive mutation strategies is proposed. By introducing an energy-aware operator, the algorithm achieves energy consumption balance within the robot swarm while optimizing the total path length. In addition, an event-triggered dynamic rescheduling mechanism is designed. When sudden robot failures or task updates are detected, a Local Greedy Insertion (LGI) strategy is activated to achieve rapid task takeover and reallocation. Experimental results show that the proposed IGA consistently reduces the system’s state of charge (SoC) range to less than 1%, significantly outperforming baseline algorithms. It strikes an excellent balance between solution accuracy and computational time overhead. Finally, by simulating sudden new tasks and robot failure scenarios, the effectiveness of the dynamic rescheduling mechanism is verified, ensuring the timeliness and high robustness of the system.
Wan et al. (Tue,) studied this question.