Locker-based Truck-Drone Routing with Integrated Considerations of Pickups, Deliveries, and No-Fly Zones

2026-06-26 · Source: Artificial Intelligence · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Robotics & Autonomous Systems, Logistics & Freight Transportation · Depth: Expert, quick

Summary

A new locker-based truck-drone routing problem, LTDRP-PDNF, addresses complex last-mile logistics by integrating parcel delivery, return pickup, battery-constrained and load-dependent drone flights, and necessary detours around no-fly zones. This problem aims to minimize the total operational cost of drone-equipped truck fleets. To solve LTDRP-PDNF, researchers developed a two-stage deep reinforcement learning-based neural heuristic. The first stage employs an attention-based encoder and a Bidirectional Gated Recurrent Unit decoder to tackle the truck-only routing problem, framed as a capacitated vehicle routing problem. The second stage then combines a policy-transfer strategy with a hybrid dispatch assignment heuristic to construct fully coordinated truck and drone routes. Experiments show this method outperforms metaheuristic and neural heuristic baselines in most cases, achieving short computation times and offering a scalable solution under practical constraints.

Key takeaway

For logistics managers evaluating advanced last-mile delivery solutions, this research demonstrates a robust framework for integrating truck-drone operations. You should consider adopting deep reinforcement learning approaches to manage complex variables like pickups, deliveries, battery limits, and no-fly zones simultaneously. This method offers superior performance and faster computation, enabling more efficient and cost-effective fleet management in real-world, constrained environments.

Key insights

A two-stage deep reinforcement learning neural heuristic effectively solves complex locker-based truck-drone routing with integrated pickups, deliveries, and no-fly zones.

Principles

• Locker systems enhance drone delivery flexibility.
• Integrated routing optimizes complex last-mile logistics.
• DRL offers scalable solutions for operational constraints.

Method

A two-stage DRL neural heuristic. Stage one uses an attention-based encoder and Bi-GRU decoder for truck-only CVRP. Stage two applies policy-transfer and a hybrid dispatch assignment heuristic for coordinated truck-drone routes.

In practice

• Optimize truck-drone fleet operational costs.
• Manage drone battery and load dependencies.
• Plan routes around restricted airspace.

Topics

• Truck-Drone Routing
• Last-Mile Logistics
• Deep Reinforcement Learning
• Neural Heuristics
• Fleet Optimization
• No-Fly Zones

Best for: Research Scientist, AI Scientist, Machine Learning Engineer, Robotics Engineer

Related on AIssential

• AI system learns to keep warehouse robot traffic running smoothly — A hybrid AI system uses deep reinforcement learning to optimize warehouse robot traffic, achieving significant throughput gains.
• Think Like a Pilot: Fine-Grained Long-Horizon UAV Navigation — An asynchronous VLM and diffusion model architecture enables fine-grained, long-horizon UAV navigation with real-time continuous control.
• A Neuromorphic Reinforcement Learning Framework for Efficient Pathfinding in Robotic Mobile Fulfillment Systems — Neuromorphic reinforcement learning provides ultra-low-power, high-fidelity pathfinding for Robotic Mobile Fulfillment Systems.
• AirDreamer: Generalist Drone Navigation with World Models — AirDreamer enables generalist drone navigation using RL on a world model, achieving robust sim-to-real transfer in complex, unseen environments.
• The Robotics Industry and Its Android Moment — A shared, cloud-backed intelligence layer is crucial for scalable, adaptable, and continuously improving autonomous robotics.
• Scaling Self-Play for End-to-End Driving — Self-play DAgger training in a high-throughput simulator enables scalable, robust end-to-end autonomous driving from pixels.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Artificial Intelligence.