Abstract:

Objective To meet the demands for intelligent spinning workshops driven by the widespread use of automated guided vehicles (AGVs), this study addresses critical challenges in AGV path planning. For single-AGV path planning, prevalent issues include a high risk of algorithm stagnation leading to deadlock, excessive redundant turns in planned paths, and slow convergence speed. Concurrently, frequent path conflicts among multiple AGVs require effective resolution strategies. To tackle these problems, this research aims to develop an improved ant colony optimization algorithm combined with a hybrid conflict resolution strategy. The objectives are to enhance the efficiency of individual AGV path planning and achieve conflict-free path coordination for multi-AGV collaborative operation. This integrated approach is expected to significantly improve the overall operational efficiency of AGV systems within spinning workshops.

Method This study utilizes grid-based modeling aligned with spinning workshop workflows to address path planning from both single-AGV and multi-AGV collaborative perspectives. For individual AGV path planning, the conventional ant colony algorithm was enhanced through three key innovations; i. implementing a path backtracking mechanism to resolve diverse deadlock scenarios, which actively extracts trapped agents upon deadlock detection; ii. developing a pheromone reward-penalty strategy to evaluate each iteration's solution paths and apply corresponding rewards/penalties; iii. introducing a turning minimization heuristic that guides route selection during iterations through orientation-based optimization in the heuristic function. For multi-AGV conflict resolution, a hybrid priority-time window algorithm was proposed, featuring the priority-based task allocation combined with time window conflict detection to classify conflicts, visual identification of conflict zones through time window rearrangement on conflict segments, and customized resolution strategies for different conflict types (node contention, head-on collisions, and so on). The system iteratively executes this detection-resolution cycle until eliminating all conflicts.

Results Experimental validation confirmed the enhanced algorithm's superior path planning performance for individual AGVs across both simple and complex environments. Comparative analysis against conventional and reference algorithms demonstrated that the improved ant colony optimization consistently identifies the shortest path during each iteration under simple conditions. Notably, deadlock occurrence decreased to 24.8% of conventional algorithm levels (representing a 5.9% reduction versus the reference algorithm), while convergence iterations reduced to 39.7% of traditional method requirements (a 41.3% improvement over the reference approach). The solution achieved significantly faster convergence in average path length compared to the baseline methods. In complex environmental testing, deadlock incidence diminished to 16.4% of benchmark algorithm performance (52.4% lower than the reference standard) with a 60.7% reduction in convergence iterations compared to the reference algorithm. Comprehensive path data analysis confirmed global optimization in turning metrics. For multi-AGV conflict resolution, experimental implementation within the spinning workshop model successfully validated the hybrid strategy. Post-planning graphical analysis precisely identified and categorized three conflict types, i.e., nodal contention, pursuit interference, and head-on collisions. Application of minimum-cost resolution strategies for each category eliminated all conflicts, with temporal analysis of pre- and post-resolution time windows confirming zero repeated node occupancy. These results jointly verify the effectiveness of the algorithm in achieving the dual goals of the efficiency of a single AGV and the coordination of multiple AGVs without conflicts.

Conclusion This research establishes a coordinated path planning framework for multiple automated guided vehicles in spinning workshops, integrating enhanced route planning with dynamic conflict resolution. Experimental validation confirms the improved ant colony algorithm's efficacy, where the path backtracking strategy reduces deadlocked agents to levels below 24.8% of conventional approaches, significantly strengthening global search stability particularly in complex environments. Concurrently, the pheromone reward-penalty mechanism accelerates convergence, evidenced by the solution requiring only 27.2% of traditional iterations under challenging conditions. Further optimization through heuristic function modifications yields paths with minimal travel distance and optimal turning frequency, demonstrating robust adaptability across operational scenarios. The hybrid conflict resolution system complements these advancements by accurately classifying conflict types including nodal contention, pursuit interference, and head-on collisions and implementing context-appropriate resolution strategies. Multi-AGV operational testing confirms uninterrupted workflow maintenance through effective conflict mitigation.Future work will focus on integrating production scheduling systems to optimize multi-AGV coordination and exploring computationally efficient technologies to reduce algorithmic complexity.

Key words: improved ant colony optimization, multi-automated guided vehicles conflict strategy, collaborative path planning, spinning workshop, path conflict