Research On Key Technologies For Mobile Robot Navigation Based On LiDAR

With the continued promotion of the "Made in China 2025" strategy,the world has officially entered the "era of intelligent industrial development".Intelligent mobile robots with autonomous navigation capabilities are currently a hot topic of investigation in mobile robotics.Autonomous mobile robot algorithms can be divided into five main parts: positioning,sensing,planning,control and execution.Due to its own characteristics and technical constraints,it is difficult for a single sensor to meet the needs of practical applications.The fusion of multisource sensor information can effectively improve the accuracy of spatial position,enhance the stability of data and effectively establish high precision spatial position information.It can plan its own movement route by sensing its surroundings and its own position and other information.Mobile robots are highly susceptible to path deviations during movement,and by building realtime maps to position them for path planning,they can ensure that the robot reaches its target accurately.In this paper,an all-round mobile robot experimental platform is developed using the Mc Namee wheel to investigate simultaneous localisation and map building algorithms and path planning navigation algorithms based on LIDAR multi-sensor fusion.The main work carried out is as follows:(1)An omnidirectional mobile platform was built on the basis of the Mc Namee wheel,the drive system and the drive hardware design for the relevant sensors were completed based on the STM32,and the kinematic model,chassis control and environment sensing system of the Mc Namee wheel mobile robot were analysed to establish its omnidirectional motion model;(2)This paper takes GNSS/INS/LiDAR joint navigation as the research object,and based on the optimal theory of Pose Graph,GNSS is used as the global a priori constraint of PoseGraph,and the key frame pose in Pose-Graph is utilized to optimize it to obtain a higher accuracy map.A new NDT matching localisation method relying only on LIDAR and the global point cloud map is proposed to improve its localisation accuracy in the region of weak GNSS signals through the established high accuracy point cloud map;(3)The more mainstream path planning algorithms in the field of mobile robot navigation are studied,and the intelligent path planning algorithm RRT is selected as the global path planning algorithm in this paper for its better characteristics in terms of real time and is simulated and analyzed in Matlab.In order to address the shortcomings of the artificial potential field method in terms of local unreachability,an improved repulsive function is chosen to optimize the local path planning algorithm APF.Finally,a simulation environment is built on the Gazebo platform to achieve autonomous navigation of a mobile robot and the method is validated.