Laser Vision Fusion Based on Unscented Kalman Filtering for Pose Estimation of Indoor Mobile Robot

Abstract

In the mobile robot location and navigation sys-tem, accurate pose estimation is the premise of the normal operation of the robot, so the pose estimation of the mobile robot is very crucial. Based on the unscented Kalman filter(UKF), this paper proposes a multi-sensor of two-dimensional laser rangefinder, depth camera, inertial measurement unit(IMU), motor encoder pose estimation fusion based on the kinematic model of the odometer system. By building a mobile robot pose estimation test platform, the experiment verifies that the multi-sensor pose estimation fusion based on unscented Kalman filter has better correction effect than extended Kalman filter(EKF). The proposed method effectively improves the pose estimation accuracy and enhances the environmental anti-interference ability of the mobile robot.