Research On Self-Localization Method Of The Wheeled Mobile Robot

Self-localization, as a significant research area of robotics, is the crucial step towards autonomous navigation of mobile robots, and so the research on the method to improve the position precision of mobile robot is very important. Technologies related to relative positioning of the wheeled mobile robot under the indoor environment are focused on in this thesis, and the main research work and conclusions are as follows,(1) The wheeled mobile robot experimental platform is built. The experimental results show that the platform not only can be satisfied requires from all experiments designed in the thesis but also has better robustness. And then the error model of the platform is introduced to analyze the dead reckoning.(2) The factors of effects on system error from the unequal wheel diameter, uncertainty about the wheelbase are analyzed, and then the UMBmark experiments are carried out based on the error model. The results show that UMBmark Calibration is effective to correct the wheel diameters and the wheelbase, and corrections on system parameters can enhance the position-estimation accuracy of mobile robot greatly.(3) The wheel-slippage model is proposed on the basis of the measurement information from the encoders and the MEMS gyroscope when the wheeled mobile robot performs straight-line motion, and then the discrimination as well as the calibration on the wheel-slippage is derived. The experiment results show that the wheel-slippage can be identified in time, and the localization accuracy can be improved effectively by calibrating wheel-slippage.(4) The matches between laser radar scanning data and map are realized by g-weighted Hough transform and Effective Area of Plane (EAP), which is improved efficiency of processing laser scanning data greatly. Weighted least squares are employed to calibrate orientation error and position error based on the matches. The experiment results show that the point-to-point least squares has better calibration effect on the self-localization.(5) Extended Kalman Filter (EKF) is implemented to fuse the information from encoders and laser radar to locate mobile robot on the basis of the data fusion technology. The experiment results show that orientation error and position error can be corrected effectively and mobile robot’s pose is also tracked accurately by EKF.