Research On Pose Estimation And Compensation And Velocity Interpolation Of Mobile Robot

Mobile robot is a research hotspot in robot industry at present.With the popularization of costs,their applications are bound to show explosive growth.However,due to different application scenarios and fields,different robots devices have significant difference in drive mode and control system.As a result,the generality of algorithms and reusability of the codes are restricted.Therefor,based on the consideration of good universality,this paper begins with the description of the running state of mobile robots in space,and aims to design a positioning system framework that can configure sensors and algorithms flexibly.On this basis,the article focuses on the working principle of every sensor and its data calibration method,studying its use characteristic and error compensation indepth.Then,methods of pose estimation using sensor data are studied aiming to explore the advantages and disadvantages of different algorithms.Taking into account the poor performance of the actual movement of the mobile robot,its speed interpolation methods are studied to enhance the smoothness of the movement.Finally,these functions are implemented in combination with ROS that with good code reusability.Firstly,based on the previous research results of the research group,this paper studied the hardware configuration technology of mobile robots,proposed and completed the hardware upgrade of the existing mobile robot platform,analyzed the hardware configuration before and after the platform upgrade in detail,and studied the control system software.The method of porting the control system software between different platforms is studied,and the software migration of the platform after the hardware upgrade is realized involving the control system and the man-machine operation interface.When constructing the control system,the JNPF-4WD series of mobile robots are mainly integrated,and the working principle and data processing method of the sensors equipped are studied,which lays the foundation for the subsequent research on pose estimation.Secondly,the methods of pose estimation using sensors equipped with JNPF-4WD series mobile robots are studied emphatically.Stochastic gradient descent method is used to fuse and compensate the original data of IMU module.The modeling and processing process of the fusion algorithm is studied,and the data can be directly used by the control system after processing.The principle of pose estimation for particle filter is studied,which provides the basis for selecting the appropriate navigation algorithm based on the actual hardware configuration.The method of pose estimation using two-dimensional laser radar is studied,and the pose estimation is performed using Gauss-Newton iteration method,which has high calculation effiency.Finally,the motion state of the mobile robot is described in a 15-dimensional space,an EKF-based pose fusion algorithm is established,and a positioning system framework that can configure sensors and algorithms flexibly is designed.Besides,based on the actual motion performance of the mobile robot,the relevant velocity interpolation algorithms are studied.Based on the research of local path planning algorithm,a speed interpolation method based on similar triangles is adopted in navigation,and the acceleration upper limit is scaled to ensure that the motion trajectory remains unchanged after the speed interpolation;in manual operation,the S-type model is proposed to avoid the sudden start-stop phenomenon of mobile robots.Finally,the above studies are verified through experiments,including map construction of mobile robots,positioning effects under different sensor and algorithm configurations,and movement performance after speed interpolation.Based on the existing four-wheel differentially driven mobile robots,this paper focuses on the methods for estimating and compensating the pose of various sensors,and has done a lot of experimental verification on the proposed positioning system framework.The experimental results show that the framework is flexible and versatile.The velocity interpolation of differentially driven mobile robots is studied to improve the slippage and unsteady motion during the movement.The experimental results show that the mobile robot can achieve a smooth movement with excellent performance after the interpolation.