Design And Implementation Of Multi-sensor Information Fusion Autonomous Navigation Control System Based On ROS

In recent years,robots have begun to enter commercial use in many fields,and the related core technologies of robots have also developed rapidly.Among them,indoor service robots,as one of the important application scenarios of robots,have begun to play a role in catering,education and logistics.As one of the core technologies of robots,autonomous navigation control technology has gradually matured.In order to solve the indoor scene application,the traditional lidar-based singlesensor autonomous navigation control system can not adapt to the complex environment.The robot autonomous navigation control system based on multi-sensor information fusion has become an hotspot.This paper uses Kinect v2,rplidar sensor,and uses C++ programming language and open source software libraries such as OpenCV,Kinect SDK and gmapping in ROS system platform to design an indoor robot autonomous navigation control system.The system uses multiple sensors to collect data,optimizes the raw data and uses the data fusion technology to process the sensor data to obtain comprehensive data.The integrated data is transmitted to the navigation control module to control the robot to move to the target by the path planning algorithm.This paper first analyzes the requirements of the system,then gives the overall framework design of the system,and briefly introduces the depth image repair module,data fusion module and navigation control module.Secondly,the depth image restoration module is designed based on pixel filtering and median filtering to repair the noise and information invalid areas in the original depth image of Kinect v2,and improve the accuracy of information collection in the robot operating environment.Thirdly,for the indoor complex environment,this paper proposes to use the Kinect v2 depth image data to complete the data precision optimization and three-dimensional information supplementation to obtain the fusion data through the adaptive weighting algorithm.Then based on the fused data,the navigation control module obtains the environmental data and completes the global path planning using the improved A* algorithm.In the process of robot operation,the artificial potential field method is used to perform dynamic obstacle avoidance.The problem of target unreachable and local minimum point of artificial potential field method is solved by modifying the repulsion field function and temporary target point.Finally,the paper tests the important modules and systems of the system.The results show that the system has achieved the expected design goals and provided the basis for the actual system application of the robot.