Research On Localization And Obstacle Detection And Measurement Of Indoor Mobile Robot Based On Multi Sensor Fusion

At present,the autonomous control performance of mobile robot continues to improve.To achieve autonomous and safe movement of robot,localization technology and obstacle detection technology are the two most critical aspects.Because there are so many problems for localization and obstacle detection in relatively complex conditions.The method which based on multi-sensor fusion has become a new trend.Therefore,this paper used indoor mobile robot as the platform,and used multi-sensor data fusion technology to explore the indoor localization and obstacle detection scheme of mobile robot.The detailed research contents of this paper are as follows:(1)A mobile robot experimental platform was built by selecting a variety of sensors and other appropriate software and hardware.The various sensors used in the experimental platform were introduced.The calibration of the camera and the joint calibration of the camera and the lidar were researched for the later sensor data fusion.(2)Aiming at the indoor localization problem of mobile robot,the advantages of IMU with high accuracy in a short time were used,and the extended kalman filter was combined to correct the accumulated error of odometer caused by wheel slip.At the same time,the modified odometer datas were used as the initial value of lidar PL-ICP algorithm,to improve the efficiency of attitude calculation.Finally,the final pose of the robot was obtained by weighted fusion of odometer and lidar localization results.The experimental results show that this method can improve the localization accuracy of mobile robot.(3)Aiming at the obstacle detection and measurement problem of mobile robot,the study used deep learning network SSD to detect environmental obstacles.For that the ground area couldn't be effectively removed during region segmentation,a color-based adaptive threshold segmentation method was proposed to perform resegmentation to remove the ground.For the extraction of obstacle contours,a contour extraction method based on lidar point feature information was proposed.Finally,a method combining the principle of perspective was proposed for measurement,for the measurement of hollow obstacles such as chairs.The feasibility of obstacle detection and measurement algorithms were verified through experiments.