Robomaster Robot Automatic Targeting System Based on YOLOv11n and DeepSORT

Abstract

Robomaster is a competitive robotics event where only hits on a robot’s four armor plates count as valid damage. To achieve fast and precise tracking of these armor plates, especially in the face of the enemy robot’s high-speed movements and the complex background of the arena, this paper proposes an automatic aiming system. The system integrates the YOLOv11n detection algorithm for target recognition and the DeepSORT tracking algorithm for continuous target tracking. The YOLOv11n model is responsible for detecting the armor plates in real-time, while DeepSORT keeps track of the target’s trajectory across multiple frames. To handle target occlusion and improve tracking accuracy, the system uses a Kalman filter to predict the robot’s movement and incorporates deep appearance features for more robust data association. This combination helps maintain precise tracking even when the target is partially blocked. Experimental results show that the proposed system significantly enhances the robot’s ability to track and hit the enemy’s armor plates, leading to improved accuracy in dynamic and challenging competition scenarios.