A Multiplayer Target Defense Game Between Quadrotor Teams

The rapid development of unmanned aerial vehicles (UAVs) has created enormous opportunities as well as emerging security risks. As increasing drone accidents and threats have been reported, the research interest in counter-UAV missions has also grown. The goal of this thesis is to find a guidance law, or a defending strategy, in a counter-UAV mission that helps a group of defenders to reach in vicinity of a group of invading drones so the latter can be intercepted before entering a target area. For the defending strategy to be robust, a conservative assumption is adopted that the invaders are also capable of optimizing the invading strategy, instead of following a predefined path. This gives rise to a target defense differential game. Because player velocities have a fundamental impact on the solution, the problem is divided into two branches, where the defenders travel faster and slower respectively. When the defenders travel faster, capture is guaranteed, therefore the focus is on how far from the target area the invaders can be captured. The proposed defending strategy is fully distributed, containing a local subgame for each defender and a task assignment problem that allocates the invaders among the defenders. The optimality of the proposed solution is extensively discussed. When the defenders travel slower, an invader can only be captured by two cooperative defenders in most situations. In addition, the defenders must recede toward the target area in exchange for a chance of capture. The optimal defending strategy is a tradeoff between the receding distance and other assisting factors, such as the cooperation from another defender, the boundary of the game region, or the topology of the target area. These three aspects are discovered and explained in the solution of three different slower-defender games. The focus of solving a slower-defender game is the defenders' winning condition that whether the invaders can be captured outside of the target area. Two methods are proposed to design a computationally effective state-feedback defending strategy to meet these winning conditions. The proposed defending strategies are validated through experiments.

Department dc:contributor.department

Aerospace Science and Engineering

Year dc:date.issued

2022

Author dc:creator

• Fu, Han

Advisor dc:contributor.advisor

• Liu, Hugh HL

• Counter-UAV
• Differential game
• Dominance region
• HJI equation
• Optimal control
• Target defense game

Statement dc:rights

• Attribution 4.0 International

Licence dc:rights.uri

• http://creativecommons.org/licenses/by/4.0/

Handle dc:identifier.uri

http://hdl.handle.net/1807/125191

OAI identifier oai:identifier

oai:utoronto.scholaris.ca:1807/125191