An Improved Artificial Potential Field Method with LiDAR for Autonomous Navigation in Dynamic Food Production Environments

The rapid development of robotics has transformed industrial automation, with significant advancements in the food industry. Autonomous navigation is critical for enhancing efficiency and flexibility in dynamic production environments, where robots transport raw materials and semi-finished products across facilities. Traditional Artificial Potential Field (APF) methods, while effective for real-time obstacle avoidance, often face challenges such as local minima, limited dynamic obstacle handling, and unsmooth paths. To address these limitations, this study proposes an improved APF method integrated with LiDAR for real-time perception and dynamic field generation. The approach incorporates global path guidance to overcome local minima, Kalman filtering for dynamic obstacle state prediction, and Bezier curve smoothing to enhance path continuity and stability.Simulation experiments conducted in a food production environment demonstrated significant performance improvements. The proposed method achieved a 5.6-meter average path length, 97% obstacle avoidance success rate, and 55 ms planning time, outperforming the traditional APF method. These results validate the enhanced adaptability, efficiency, and safety of the improved APF in dynamic environments. This study provides a robust and efficient navigation solution for intelligent food manufacturing, contributing to the broader adoption of autonomous systems in industrial applications.