Research On The Positioning Of Indoor Mobile Robot Based On LiDAR/INS Integrated Navgation

High resolution positioning by multi-sensor fusion is the important technology to achieve intelligent mobile robot.Compared with outdoor working robots indoor working robots have more stringent requirements on positioning accuracy that is difficult to achieve.In this paper,a self-designed Mecanum wheeled omnidirectional mobile robot was adopted as the research platform,and LiDAR and INS sensors were used to research the integrated positioning algorithm in the indoor environment.Firstly,the sensor used in combination positioning is modeled and the processing method of sensor data is given.That includes the definition of coordinate system adopted,the analysis of the working principle of LiDAR,the optimization algorithm of sensor data noise using S-G filter and the algorithm of motion distortion removal aided by IMU.The INS navigation update calculation method and the principle of system error generation are given,and the system error equation is established.Secondly,based on the principle of omnidirectional motion of Mecanum wheel,kinematics equations related to control are given.And the robot hardware and software design are introduced respectively,motor drive circuit control circuit and the inertial navigation system circuit is given of device selection and hardware design.Scheme of software design are mainly introduced based on ROS operating system software architecture design and embedded software design scheme based on FreeRTOS.Thirdly,the scanning matching principle based on ICP method and NDT method is analyzed and introduced,and the calculation process of the two methods is given,and the K-D tree index method is given to improve the matching efficiency of the nearest neighbor points for ICP method.Introduce Kalman multi-sensor fusion algorithm adopted in this paper,analyzes the principle of KF and EKF data fusion,establishes a state and measurement model for the combination of LiDAR and INS,and designs a loosely-coupled combined positioning filter.Finally,the robot experiment platform designed in this paper was used to test ICP algorithm and NDT algorithm respectively in Gazebo simulation environment and physical turntable environment,and the matching accuracy and computational efficiency of the two algorithms were verified.The comprehensive analysis of the computational efficiency of NDT algorithm is far better than ICP algorithm and the actuarial accuracy can meet the requirements.Experiments were carried out on pure LiDAR matching positioning and combined positioning in indoor environment,respectively,to verify the positioning accuracy of rectangular and circular trajectories in linear and curvilinear motion in pre-calibrated environment.Experimental results show that the combined positioning accuracy is much better than the uncombined positioning accuracy,and the combined positioning RMS error is 0.072 m and 0.025 m respectively when the line and curve move.The accuracy can meet the requirement of indoor autonomous function of robot and has certain practical value.