Hippopotamus Optimization Algorithm Applied To Dynamic Vehicle Routing In Quick Commerce

Classified as NP-hard, the Quick Commerce Dynamic Ve- hicle Routing Problem (QC-DVRP) can benefit from bio-inspired opti- mization techniques for last-mile logistics. This study applies the Hip- popotamus Optimization Algorithm (HO), a recent metaheuristic that requires no algorithm-specific hyperparameters and, to the best of our knowledge,hasnotbeenpreviouslyevaluatedonroutingproblems,tothe QC-DVRP. It is evaluated against five metaheuristics (PSO, GA, GWO, SSA, GTO) in a discrete-event simulator with Poisson demand (λ= 5.0 orders/min) and re-optimization every 5 minutes. Over 30 independent runs, the Friedman test with Nemenyi post-hoc (α = 0.05, CD = 1.38) identifies two performance groups: HO, with an average rank of 1.23 in Fitness, ranks statistically above all five competitors (∆ ≥1.94, all > CD), which form a single equivalent group in Fitness. HO achieves the best delivery success rate (DSR = 75.9% vs. 74.6–75.1%), the best workload balance (WBI = 889), and the best overall fitness (Z = 6527), with no significant differences in delivery time (ADT, p= 0.355). While the absolute improvements are modest (e.g.,∼1.2 percentage points in DSR), they are consistent across metrics and statistically significant. The results support HO’s competitiveness hypothesis, suggesting that its design without algorithm-specific hyperparameters and three sequen- tial phases may offer a favorable exploration-exploitation balance for periodic re-optimization in dynamic scenarios.