Real-Time Cooperative Trajectory Optimization for 3D Infrastructure Inspection in Mixed Indoor-Outdoor Environments
Antonis Nikolaidis, Antreas Anastasiou, Christos Laoudias, Panayiotis Kolios · 2025
This paper presents a real-time cooperative trajectory optimization framework (RECON) for autonomous 3D infrastructure inspection using heterogeneous UAV teams. RECON addresses GPS-denied navigation and intermittent communication challenges through a two-stage approach: (1) LiDAR-based volumetric mapping via explorer UAVs, and (2) decentralized path planning for image collection (photographer) UAVs using a multi-Traveling Salesman Problem (mTSP) formulation. The framework integrates OctoMap for environment discretization and evaluates three communication models to ensure robustness under varying connectivity. Simulations across indoor-outdoor scenarios—including urban structures and industrial facilities—demonstrate the progressive improvement in average observation quality ($q_{\zeta}$) over time. Results confirm RECON's ability to prioritize regions of interest while minimizing redundant exploration, even under strict communication constraints. The framework advances scalable solutions for infrastructure monitoring by enabling adaptive coordination in dynamic environments with limited sensing capabilities.