Design And Implementation Of An Autonomous Navigation System For Mobile Robots Based On Multi-Sensor Fusion

In the development of modern society,mobile robots have been widely used in various aspects of production and life,which has also led to increasingly higher requirements for the performance and functions of mobile robots.Nowadays,most mobile robots rely on preestablished 2D grid maps to achieve autonomous navigation in unknown indoor environments.However,they cannot accurately and timely avoid dynamic and static obstacles that do not exist in the prior environment maps.Based on multi-sensor fusion technology,autonomous navigation of mobile robots has become a research hotspot among many scholars.In this study,a mobile robot autonomous navigation experimental platform was built based on a single-line laser radar,RGB camera,depth camera,and wheel encoder.The platform starts with the algorithms of environmental perception,real-time localization and mapping,and autonomous navigation,and achieves autonomous navigation of mobile robots in unknown enclosed indoor environments.The main research contents are as follows:Firstly,the overall goal function of the mobile robot is analyzed,and the corresponding goal function of autonomous navigation of the mobile robot is established.The overall framework of the entire autonomous navigation system is clarified.The hardware composition and software design of the system are detailed.Secondly,research is conducted on the multi-sensor fusion-based environmental perception algorithm for mobile robots.Firstly,the intrinsic and extrinsic parameters of multiple sensors are calibrated to obtain the internal parameters and relative positions of the sensors.Dynamic and static obstacles are detected using multiple sensors.RGB camera combined with Yolov7-tiny algorithm is used for detecting human dynamic obstacles,and the depth camera is used for screening and detecting low convex static obstacles on the ground.Afterwards,a multi-sensor fusion-based instant positioning map building function is studied,using upstream environmental sensing information combined with the original laser scanning information for environmental information fusion,and using odometry and point cloud map matching information using extended Kalman filter fusion for positional update positioning,and finally obtaining a two-dimensional raster map.Finally,research and analysis are conducted on the path planning of autonomous navigation function.The A-Star algorithm is optimized based on an improved heuristic function to improve its search efficiency in large enclosed indoor environments.A new evaluation function is designed for local path planning that is insensitive to dynamic and static obstacles to improve the dynamic window algorithm,thus achieving the autonomous navigation function of the mobile robot.In the experimental part,a combination of simulation software and physical verification is used to place the mobile robot with multi-sensor fusion function in an unknown scenario.The mobile robot can realize environment perception,instant localization and map building and autonomous navigation functions,and the robustness of this mobile robot autonomous navigation system is verified.