Reaserch On Cooperative Localization Of Multiple Mobile Robot System

The multiple mobile robot system has the in comparable advantages and potentials compared with the sigle mobile robot.Firstly,the multiple mobile robot system can achieve the goal by cooperation which cannot be achieved by single mobile robot.Secondly,the efficiency will be improved considerably through the cooperation of the multiple mobile robots.When the working environment changed or the some of the mobile robots failure,the multiple mobile robot system will still finish the task by cooperation.Since there are so many advantages,the multiple mobile robot system has become an important developing direction of robotics since 1970 s.This thesis focuses on the enhancement of the localization accuracy of the multi-mobile robot system and the research content includes:The core problem of the mobile robot localization is information fusion.Total observable of the system is the premise of the information fusion.As a result,the observability analysis is firstly accomplished.In order to solve the problem of the observation analysis of the multiple mobile robot system,an observation analysis method based on the relative position measurement graph(RPMG)is proposed in the thesis.The RPMG observation analysis method combines the relative properties of gragh theory and the observation method of nonlinear system based on Lie derivative together.The total observable condition of any-mobile-robots and any-landmarks can be proved.Accurate measurement information is the basic prerequisite for the multi-sensor information fusion and the mobile robot localization.It is well known that the observation information is obtained from the stereo camera or monocular camera(only the monocular camera is considered in this thesis),which means the observation information is extracted from the image information.As a result,the one of the key research problems is how to extract the observation information from the image sequences.In order to solve the problem of the horrible result if there is a big variance difference between the target area and the background area in the process of image segmentation,a modified Otsu algorithm is proposed in this thesis.In order to solve the conflict between the ability of denoising and the ability of detail information extracting in the image edge detection process,a modified LoG algorithm is proposed in this thesis.Above all,a modified image edge detection algorithm is proposed and the effectiveness of novel algorithm is verified by the actual images.State estimation algorithm is the effective method of the information fusion.Consider the characteristics of the mobile robot,the state estimation method is researched in the thesis.Since the measurement model is uncertain,the idea of the particle filter based on the probability theory is introduced in the belief function filter and the particle filter algorithm based on belief function is proposed in the thesis.The real data from University of Toronto Institute for Aerospace Studies(UTIAS)is used to verify the effectiveness of the algorithm.The experiment results show that the positioning error of the multiple mobile robot system is reduced by utilizing the algorithm which proposed in the thesis.As a result,it is proved that the algorithm proposed in the thesis can be used to enhance the positioning and localization accurarcy of the multiple mobile robot system.For the multiple mobile robot system which without explicit target or moving strategy,the positioning error of the mobile robot will be reduced by planning the path of the mobile robot.As a result,the path planning strategy of the multiple mobile robot system is focused in the thesis.Firstly,the information theory and entropy theory are introduced briefly and the optimal algorithm based on the information theory is also introduced.Considering the shortcoming of the control strategy of one-step-look-ahead and multi-step-look-ahead,a motion control strategy based on MPC for multiple mobile robot system is proposed in the thesis.The comparison of MPC motion control strategy and one-step-look-ahead motion control strategy is done by MATLAB.The simulation result shows that the position accurarcy of multiple mobile robot system can be enhanced by adopting a reasonable and effective motion control strategy without changing the sensors embedded on the mobile robot.