| Generally,the pose of indoor mobile robot can be gained employing odometer in conjunction with laser radar.However,it is difficult to determine the odometer model of the skid-steer robots due to its special mechanical structure and sliding between wheels and the ground during cornering.In this thesis,after the kinematic analysis of some skid-steer vehicles,a skid-steer wheeled robot is designed for study on the localization method for skid-steer wheeled robots.The odometer model of this robot is established and some research on localization using this odometer model combined with the laser scanner is finished.The main work and conclusion are as followed:(1)The kinematics analysis of the center-syimmetric and left-right symmetrical skid-steer robot is carried out,and the corresponding kinematic model is obtained.On this basis,the odometer model is derived and the method of determining error of the odometer model is proposed.(2)Compared with the advantages and disadvantages of the structures of some skid-steer robots at home and abroad,a skid-steer robot with four-wheel symmetrical structure is designed,whose relative position tolerance of the four wheels is less than 1 mm.A STM32 microprocessor with ARM cortex core is used as lower computer,which is easy to realize real-time control.CAN bus is employed for simplifying wiring and improving reliability.The built-up laser scanner is helpful for localization and obstacle avoidance.It is showed with the prototype that this skid-steer robot can meet the requirement for the study on localization method of skid-steer wheeled robots.(3)Using ADAMS and MATLAB/Simulink,the odometer model parameters and errors of the skid-steer wheeled robot designed in this thesis are determined.Experimental verification is carried out.The results show that the odometer model and its error can effectively predict the trajectory of the robot.The KNN(k-Nearest Neighbors)algorithm is applied to identify the road surface,whose eigenvalues are the left and right wheel speeds and the slip deviation values under different road surfaces.The results show that the algorithm can better identify the road surfaces.(4)The particle swarm algorithm is used to simulate the skid-steer wheeled robot designed in this thesis.It is verified that the algorithm can estimate the position.Meanwhile,the algorithm is modified.Combined with the experimental environment map and odometer model obtained by laser radar and its error,the particle positioning algorithm was used to carry out the positioning experiment of the robot.The experimental results show that the modified particle filter positioning method can better estimate the position of the designed skid-steer wheeled robot. |
PDF Full Text Request