Visual area coverage with collaborative autonomous aerial vehicles

• Κύριος συγγραφέας: Μπούσιας, Νικόλαος
• Άλλοι συγγραφείς: Δερματάς, Ευάγγελος
• Μορφή: Thesis
• Γλώσσα: English
• Έκδοση: 2019
• Θέματα: Autonomous swarms; Collaborative robotics; Area coverage; Decentralized control; Muti-agent systems; Sensor networks; Εναέρια ρομπότ; Οπτικοί αισθητήρες; 629.892 637

The current thesis examines the problem of visual surveillance of a planar, convex region of interest by a fully autonomous, heterogeneous swarm of mobile aerial vehicles, equipped with Pan-Tilt-Zoom cameras of conical field of view and operating under GPS-induced 3D localization uncertainty. The proposed gradient-based collaborative control framework, serv- ing as trajectory generator, utilizes a computationally less strenuous Voronoi-free tessellation strategy to monotonically optimize the team’s collective visual coverage over the region of interest, both in terms of total monitored area and the quality of the acquired information. The designed distributed controllers of the flock consist of a multilevel structure, featuring inter- agent cooperation and system-specific trajectory tracking, thus offering significant flexibility regarding the physical implementation of the swarm. For the purposes of this thesis, a swarm of quadrotors is assumed, the dynamics of which are examined and an optimal, dual level (posture-orientation) controller for path following is designed. Multiple simulations where conducted in order to evaluate the efficiency and robustness/resilience of the proposed scheme in a variety of configurations. The required range of communications between the robots of the system is examined and a multi-hop, deterministic communication algorithm is offered.