Accurate positioning and precise map construction are fundamental for the successful completion of inspection tasks by cable trench inspection robots.Traditional laser SLAM algorithms often encounter challenges such as the'long corridor effect'and z-axis drift in cable trench environments.To address these issues,an improved EKF_LOAM algorithm,utilizing multi-sensor fusion,is proposed.By extending the Kalman filter(EKF)to integrate wheel odometry and inertial odometry into the LeGO_LOAM framework,the'long corridor effect'is suppressed.Additionally,based on the EKF_LOAM algorithm,the Z-axis velocity measurement of IMU is introduced,and the adaptive covariance equation based on the number of feature points is designed to constrain the Z-axis drift caused by the serious absence of surface feature points.Based on the EKF_LOAM algorithm,the IMU measurements of z-axis linear velocity is introduced,and the adaptive covariance equation based on the number of feature points is designed to constrain the z-axis drift caused by significant feature point loss.Experimental results conducted in both simulated and real cable trench environments demonstrate that the proposed algorithm reduces the final positioning error by over 40%compared to the LeGO_LOAM algorithm and by 4.57%compared to the Cartographer algorithm.Furthermore,the z-axis direction error is reduced by 49%compared to the EKF_LOAM algorithm,indicating the superiority of the proposed algorithm in cable trench environments over traditional LeGO_LOAM,Cartographer,and EKF_LOAM algorithms,making it more suitable for cable trench inspection tasks.
NETL
NSTL
维普
万方数据
