Research On SLAM For Rescue Robots Based On Multimode Fusion

Rescue robots are increasingly used in various disaster rescue conditions,however there are still many deficiencies in the autonomous navigation of robots in complex environments with weak textures.The lidar sensing information is relatively simple,therefore the environmental map is not constructed abundantly enough.Although visual cameras can obtain the richer environmental information,it is easily affected by sunlight and darkness.Its stability and reliability are not so good that the rescue robots can not adapt to complex environments.In this paper,a SLAM navigation algorithm suitable for rescue robots is proposed by fusing the multi-modal sensing information of lidar,vision camera and IMU.It enables rescue robots to continuously sense location information robustly and accurately in various challenging environments.The main research work of this paper is summarized as follows:(1)In order to improve the positioning accuracy,the lidar,visual camera and IMU are fused for calibration.The multi-mode internal parameters are transformed into the coordinate system to obtain the external parameters.Then the Autoware algorithm is used to fuse the transformation matrix into unified coordinates.Finally,the fusion calibration algorithm is verified effectively by experiment,which provides a basis for SLAM navigation research of rescue robots.(2)Aiming at the problems of easy loss of point features and matching failure in weakly textured environments with similar structures,a PL-VIO point-line combination visual-inertial odometry feature extraction algorithm is innovatively proposed based on the LVI-SAM algorithm framework of 3D lidar,visual camera and IMU fusion.The feature extraction algorithm increases feature diversity and improves feature matching accuracy.The sliding window method is used to optimize the measurement residuals of points and lines and the IMU measurement residuals,which improves the performance of the pose estimation of the LVI-SAM algorithm.It provides more accurate and reliable rescue environment information for rescue robot navigation.(3)The KITTI outdoor standard data set is used for evaluation in order to verify the performance of the algorithm in this paper.Then,a software and hardware verification platform for the SLAM navigation algorithm of the rescue robot is built based on the Alman wheeled mobile robot.Through the outdoor scene test experiment,our proposed algorithm in this paper is compared with the LVI-SAM algorithm,and the mapping effect and positioning error are analyzed.The results show that our algorithm is more superior for the complex environment of rescue robots.