Experimental Demonstration of Optimized Single Target Tracking with Reinforcement Learning

This paper presents the first experimental demonstration of optimized single target tracking using reinforcement learning methods in a radar network. Recent studies have shown the effectiveness of such methods in real-time optimization of sensor settings for target tracking. In this research, we integrate an optimization framework based on reinforcement learning methods into an existing testbed comprising multiple X-band monostatic radar nodes controlled by a centralized processing unit. Using two experimental scenarios in different environments and with distinct optimization goals, we illustrate the real-time adaptation and optimization capabilities of the radar network. The results demonstrate improved tracking accuracy and resource management, highlighting the potential of reinforcement learning for cognitive radar applications.