An Adaptive Ant Colony System for Skip Loader Operations in Roll-on/Roll-off Logistics

Skip loader operations in Roll-on/Roll-off logistics involve alternately transporting empty and full waste containers under strict constraints related to vehicle capacity, container compatibility, stackability, and service time windows. These operations require careful coordination of skip handling and route planning to ensure feasibility and efficiency. This paper presents an adaptive Ant Colony Optimization (ACO) system tailored to the unique demands of skip loader logistics. Each service request is modeled as a sequence of up to five atomic operations: loading or unloading empty or full skip, and executing skip swaps. These modular atomic operations enable flexible modeling of diverse operational scenarios, including multi skip stacking and varying disposal or delivery points. Ants construct routes incrementally while maintaining an internal state that tracks vehicle load configurations and skip statuses. This state-aware routing mechanism ensures that all transitions are feasible, avoiding deadlocks and invalid loading states. Routing decisions are guided by a composite heuristic that integrates travel cost, service duration, stacking feasibility, and time window alignment. Pheromone trails are reinforced both on spatial paths and action transitions, guiding the search toward combinations of actions that have previously led to feasible and efficient solutions. The system was evaluated on extensive operational data from municipal waste collection fleets, featuring multiple skip loader types, heterogeneous skip inventories, and realistic time-window constraints. The results show that the approach consistently produces feasible high-quality routes, maintains valid load states, utilizes the vehicle capacity effectively, and converges rapidly. The proposed adaptive ACO framework offers a robust and extensible foundation for solving Roll-on/Roll-off routing challenges and supports modular service planning and the integration of operational constraints in complex skip handling environments.