Researchers from the School of Artificial Intelligence and Automation at Huazhong University of Science and Technology have developed a novel approach to enhance multi-target tracking capabilities for unmanned aerial vehicles (UAVs) operating in jamming environments. The study, led by Ziang Wang, Lei Wang, Qi Yi, and Yimin Liu, addresses a critical challenge in modern military operations: maintaining accurate tracking of multiple targets despite the presence of electronic jamming.
UAVs have become indispensable in both military and civilian applications, with multi-target tracking being one of their most critical functions. However, the effectiveness of radar-based tracking can be severely degraded by jammers, which disrupt radar signals and compromise operational accuracy. To mitigate this issue, the researchers propose a system where a swarm of UAVs dynamically switches between active and passive radar modes to track multiple targets, some of which may be equipped with jammers.
The study introduces an optimization problem to determine the most effective tracking strategy under jamming conditions. The researchers demonstrate that solving this problem is computationally complex, requiring advanced algorithms to achieve real-time decision-making. To tackle this challenge, they employ a multi-agent reinforcement learning (MARL) algorithm, which enables UAVs to learn and adapt their tracking strategies based on environmental feedback.
In addition to the MARL approach, the researchers integrate a simulated annealing algorithm to prevent UAV actions from violating operational constraints. Simulated annealing helps refine the solution space, ensuring that the UAVs’ actions remain feasible and effective even in dynamic and adversarial environments. This dual-algorithm approach enhances the robustness and reliability of the tracking system.
Simulation experiments conducted by the team validate the effectiveness of the proposed algorithm. The results show that the UAV swarm can successfully track multiple targets while adapting to jamming conditions, demonstrating significant improvements in tracking accuracy and resilience. This research highlights the potential of combining MARL with simulated annealing to address complex control problems in defence and security applications.
The findings have practical implications for modern military operations, where UAVs are increasingly deployed in contested environments. By enabling UAVs to switch between active and passive radar modes intelligently, the proposed system enhances their ability to track targets accurately, even in the presence of sophisticated jamming techniques. This advancement could lead to more effective surveillance, reconnaissance, and combat operations, ultimately strengthening military capabilities in an evolving threat landscape. Read the original research paper here.