An improved localization method based on the fusion of Kinect and odometry

Both the kinect and odometry sensor are low cost and popularly installed in the mobile robot for indoor applications, but they are suffering from low precision and are prone to error accumulation issues. To improve the localization accuracy of the mobile robot, an improved positioning method based on the fusion of kinect and odometry is proposed, which uses the Generalized Iterative Closest Point (GICP) algorithm to solve the rotation and translation. Compared with the standard Iterative Closest Point (ICP) algorithm, the improved GICP is more efficient. And with framework of the extended Kalman filter (EKF), the output of kinect is taken as the observation, and the output of odometry sensor is used as the system model. By making use of the fusion strategy, the noise is greatly suppressed. The experimental results show that the positioning accuracy and stability are improved by the proposed method.