Cloud–Edge Collaborative Autonomy Architecture for BVLOS UAV Delivery Systems

Abstract

Unmanned Aerial Vehicles (UAVs) are increasingly being considered for last-mile logistics due to their potential to reduce delivery time, traffic congestion, and operational costs. However, large-scale deployment of drone delivery services requires Beyond Visual Line of Sight (BVLOS) operations, which introduce significant technical challenges related to real-time navigation, communication reliability, airspace integration, and operational safety. Traditional cloud-centric control architectures are often constrained by network latency and intermittent connectivity, limiting their suitability for safety-critical autonomous flight decisions.

This study presents a cloud–edge collaborative autonomy architecture designed to support BVLOS UAV delivery systems. The proposed framework distributes computational and decision-making tasks across three layers: onboard UAV intelligence, regional edge infrastructure, and centralized cloud services. Safety-critical operations such as perception, obstacle avoidance, and local flight adjustments are executed on edge devices located on the UAV or nearby ground nodes, while the cloud layer performs global route planning, fleet coordination, data analytics, and model training. This distributed architecture enables responsive control while maintaining centralized oversight of delivery operations.

The paper describes the architectural design, task allocation strategy, and communication framework that enable coordinated decision-making across the cloud and edge layers. The system is evaluated through simulation experiments that measure navigation responsiveness, communication latency, and delivery reliability under varying network conditions. Results indicate that the collaborative architecture reduces decision latency and improves operational resilience compared with fully centralized approaches. The findings demonstrate that integrating edge intelligence with cloud-based coordination provides a practical foundation for scalable BVLOS drone delivery networks and supports the safe expansion of autonomous aerial logistics systems.