Research on scheduling optimization for ground-air collaborative instant delivery by takeout riders and drones in the low-altitude economy scenario

Urban instant delivery platforms face operational pressure from traffic congestion, peak-hour demand imbalance, and strict delivery time limits. To examine how low-altitude delivery resources can complement ground riders, this study investigates a centralized rider-drone collaborative scheduling problem for urban instant delivery. A weighted single-objective model is developed for ground-only, drone-only, and ground-air relay delivery modes. The model incorporates order weight, drone payload and endurance, no-fly zones, delivery time windows, relay stations, heterogeneous carrier assignment, and capacity occupation, and evaluates scheduling plans by cost, duration, load balance, and on-time service rate. To solve the resulting mixed discrete–continuous problem, an elite-triggered sparrow search algorithm (ETSSA) is proposed, combining elite-triggered saltation, dual-polarity drive, and sliding-window rebirth with random-key decoding and feasibility repair. CEC2022 benchmark tests evaluate ETSSA’s continuous-search capability, and urban instant delivery simulations evaluate delivery-mode performance, baseline comparison, ablation, rider-drone ratio, weight and multidimensional sensitivity, cost–benefit, scalability, and 24-h peak-hour operation. In the 10 × 10 km 2 synthetic service area, the ground-air hybrid scheme reduces average delivery time from 32.56 to 22.38 min, reduces average cost from 24.54 to 18.28 yuan, increases the on-time rate from 71.20% to 94.21%, and reduces load CV from 0.85 to 0.49 compared with ground-only delivery. ETSSA obtains better weighted fitness and service performance than the tested general metaheuristic, Pareto-based, and scheduling-heuristic baselines under the same scalarized evaluation setting, with additional runtime caused by the added diversity and stagnation-control mechanisms. Sensitivity and cost–benefit analyses show diminishing marginal returns from drone deployment: for a 30-rider fleet, the first 5–10 drones provide the largest marginal improvement, while 30R + 10D to 30R + 25D forms a high-score range under the adopted parameters. The findings provide mechanism-oriented evidence for ground-air collaborative instant delivery.