Research On General Autonomous Navigation System Of Differential Wheeled Robot Based On Planar Pose Sensor

At present,a variety of robots are widely used in our life,among which wheeled robots are widely used and representative.The maturity of autonomous navigation technology of wheeled robots makes the chassis manufacturing industry of large mobile vehicles begin to seek change,hoping to realize the autonomous navigation of their vehicles.At present,large mobile vehicles are customized according to the demand scene,and the motion model is the same,but the motor and other key equipment are not the same.Therefore,it is a relatively new topic to study a set of general positioning and navigation system which can carry a variety of differential chassis.Based on the military project,this paper innovatively studies a set of positioning and navigation system which can be easily installed to different differential wheeled mobile chassis.Firstly,this paper studies the robot localization and map generation algorithm,including Bayesian filter,Kalman filter,extended Kalman filter and particle filter.Aiming at the shortcomings of traditional particle filter algorithm,KLD sampling is introduced to solve the problem of fixed number of particles and robot kidnapping,which can better achieve robot localization.At the same time,aiming at the RBPF-SLAM algorithm,by improving the proposed distribution and resampling process,the problem of particle degradation in the classical RBPF-SLAM model is solved theoretically.Secondly,the theoretical research on path planning in robot navigation is carried out,which mainly includes two aspects:global path planning and local path planning.This paper makes a comparative study of Dijkstra algorithm and A*algorithm,which are suitable for global path planning.At the same time,it makes a theoretical study of DWA algorithm,which is widely used.Then,it makes a simulation analysis of the algorithm expected to be carried to the chassis based on MATLAB,and obtains a path planning algorithm suitable for actual navigation.Then,gmapping algorithm based on improved RBPF algorithm is used for robot mapping analy sis.Because of impro ving the accuracy of the input prior probability,the accuracy of robot mapping can be improved.Based on this feature,this paper uses a plane pose detection device to solve the odometry,and compares the effect of traditional motor encoder and motor encoder combined with gyroscope to solve the odometry,and finally comes to the conclusion that the plane pose detection device is better for solving the odometry.Finally,the general navigation system is built.The general design of the navigation system mainly depends on two aspects.One is based on the convenience and independence of the installation of the plane pose detection device,and the other is based on the way that two singlechip microcomputers are connected in series,and then the control signal is sent to the chassis controller through the can interface.Through the autonomous navigation experiments on two differential chassis with the same motion model,it is found that the positioning and navigation system designed in this paper can be easily installed on other chassis with the same motion model while realizing autonomous navigation,and the universal design of autonomous navigation is realized,Therefore,the design of this navigation system can complete the task of autonomous navigation and expand the robot application scenarios at the same time.