The Research Of Information Fusion Technology Applied In Mobile Robot Localization

The multi-sensor information fusion technology is the detection, multi-source data, combined. and estimates, a multi-layered, multifaceted process, thus completing the uncertainty of the information, more accurate and more comprehensive understanding and description of the measured object, in order to make the right judgments, to provide the robustness of the system.In this paper, the researches work these two aspects of the multi-sensor information fusion theory and the mobile robot self-localization theory. The main work is as follows:Firstly, the thesis introduces the basic model and the structure of the multi-sensor information fusion technology and describes the study of information fusion in the field of mobile robots, but mainly discusses its application in robot tracking and localization.Secondly, the thesis establishes mobile robot kinematic mode and sensor observation model, at the same time, analysis the commonly used algorithm for multi-sensor information fusion. At the same time, in-depth analysis of the commonly used algorithm for multi-sensor information:EKF and PF based on state estimation techniques.Then, in the thesis, we analyses and designs EKF fusion model, given the EKF positioning information, writing the matlab simulation program based on the figure. Simulation results verify the effectiveness of the EKF fusion of sensor information in the mobile robot localization.Again, in connection with degeneration of the PF, the thesis puts forward based on partition resampling PF algorithm and proposes based on information fusion of multi-sensor distribute portioned particle filter algorithm which uses the partition particle filter to calculate the local sensor state estimate, and the uses the scalar weighted fusion estimate of state of the system for information integration. Simultaneously the thesis presents PEKPF. Design experiments to verify the improved particle filter estimation performance, a comparative analysis the basic EKPF and improved EKPF of PF and improved PF performance.Finally, the thesis designs a fusion model based extended filter localization model and particle filter based on partition sampling localization model, and then uses matlab software simulating the experiment model. The simulation results show that the method has good effectiveness of information technology in the field of mobile robot.