Modular Matrix Workshop Multi-AGV Collaborative Scheduling: Dynamic Retry Mechanism and Multi-Factor Energy Consumption Optimization

Abstract To address the issues of low energy efficiency and poor machine utilization caused by frequent dynamic path conflicts in multi-AGV collaborative scheduling for matrix workshops. This paper proposes a novel scheduling optimization algorithm, namely Dynamic Retry and Energy-Aware Multi-factor Scheduling (DREAMS). Firstly, a dynamic obstacle model is constructed based on the grid layout of the workshop and the Manhattan distance, combined with a spatio-temporal conflict detection mechanism. A combination optimization model of machine load balancing and battery power attenuation is established. Then, a dynamic retry mechanism is designed to avoid deadlocks through local path re-planning; Moreover, an adaptive power-speed strategy is proposed to adjust the AGV speed based on real-time battery power dynamically. Further, the multi-factor heuristic rules are introduced to optimize machine selection and reduce workshop energy consumption. Finally, the independent contributions of each module to system performance are verified by removing the retry mechanism, conflict detection, heuristic rules, and variable speed strategy, respectively. The results show that the DREAMS algorithm reduces energy consumption by 9% to 12% compared with traditional algorithms. The collaborative effect of each module reduces the standard deviation of energy consumption by 42% and significantly improves the system's robustness.