Research on Real time Obstacle Recognition and Avoidance Strategies for Mobile Robots Based on YOLOv5
Liyuan Shao · Highlights in Science Engineering and Technology · 2025
To address the issues of insufficient detection accuracy, slow response speed, and poor adaptability in real-time obstacle recognition and avoidance of mobile robots in complex environments, this paper proposes a real-time obstacle recognition and avoidance strategy algorithm for mobile robots based on YOLOv5. To begin, using the YOLOv5 network model, an image acquisition and data preprocessing module is created to obtain high-precision and low latency obstacle location and classification information. Second, to increase recognition efficiency and reduce false positives, the YOLOv5 model structure and training parameters were optimised. Again, based on YOLOv5’s real-time characteristics, an obstacle avoidance model was built that takes into account dynamic changes in the environment as well as the influence of robot mobility status, allowing the robot to better understand its surroundings. Finally, an obstacle recognition and avoidance system based on the YOLOv5 algorithm was developed. Simulation and experimental verification revealed that the algorithm presented in this research can greatly improve mobile robots' autonomous navigation capability.