In plain English
Combining multiple bio-inspired and local-search optimization strategies into a single hybrid framework can more effectively solve large-scale vehicle routing problems — where fleets must serve many customers under complex, simultaneous constraints such as time windows, capacity limits, and multiple depots.
Year: 2026
Venue: 2026 IEEE 16th Symposium on Computer Applications & Industrial Electronics (ISCAIE)
Type: conference
DOI: 10.1109/iscaie68866.2026.11576440
External link: https://doi.org/10.1109/iscaie68866.2026.11576440
Abstract
Modern logistics networks have grown large and complex enough that standard optimization methods simply cannot handle multi-dimensional Vehicle Routing Problems (VRPs) at scale. This paper proposes a Hybrid Ant Colony System-Firefly Algorithm (ACS-FA) that links the global search power of pheromone-driven ACS with the local refinement of brightness-guided FA. Unlike conventional metaheuristics, the hybrid creates a two-way feedback channel between the two algorithms, which lets it manage high-dimensional constraints distance, weight, cargo area, and hazardous material compatibility within a single framework. We benchmarked ACS-FA against GA, ACS, FA, ALNS, and DRL-based solvers across a range of routing scenarios. The hybrid consistently produced lower route costs, better load balance, and higher capacity feasibility than all baselines. ANOVA, Kruskal-Wallis, and post-hoc tests confirmed these improvements are statistically significant at p = 0.05. Computation scales well with problem size, and the model is not sensitive to parameter choices. The routing efficiency also translates to real reductions in fuel use and CO₂ output, making ACS-FA a practical option for organizations that need both performance and sustainability from their routing systems.