UAV Swarm Autonomy through Cooperative Positioning: A Unified Approach with Distributed Graph Optimization and Decentralized MPC

Autonomous flight of unmanned aerial vehicle (UAV) swarm faces challenges in relative positioning and cooperative control. Existing methods suffer from ground reliance, high computation, and communication costs. To overcome these constraints, this study proposes a relative positioning framework based on distributed graph optimization (DGO) that integrates onboard sensor measurements of relative distance, relative angle, and inertial-based state estimations. The distributed architecture enables each drone to solve localized optimization subproblems in parallel, effectively mitigating communication congestion and computational bottlenecks. Building upon the positioning results, we develop a decentralized model predictive control (DMPC) scheme with real-time capability for simultaneous swarm formation maintenance and dynamic trajectory tracking. Results demonstrate that the integrated system achieves sub-meter relative positioning accuracy while sustaining formation stability. This work provides methodological foundations for subsequent research in swarm intelligence.