Localization of coal mine rescue robots based on multi-sensor fusion

Abstract

This paper investigates the multi-sensor fusion positioning problem coal mine rescue robots to deal with the complex catastrophic underground environment subject to poor illumination and slippery road. A multi-sensor fusion positioning system is designed by combining lidar, IMU, and wheel encoder and a multi-sensor fusion positioning method is proposed based on extended Kalman filter with fuzzy confidence. By the proposed method, the wheel slip can be described and used to regulate the confidence level of the wheel odometry information in the fusion positioning system in real-time, and change the covariance matrix dynamically to improve the adaptability of the system to the subterranean environment and enhance the positioning accuracy. The experimental results show that the system is suitable for underground positioning, and the proposed positioning algorithm is of higher accuracy than the conventional single sensor positioning methods and the extended Kalman filter fusion positioning methods.