Research on Indoor Mobile Robot Mapping based on the Fusion of LiDAR and Depth Camera

To solve the problems of low map construction accuracy, incomplete information, and poor reliability when indoor mobile robots perform simultaneous localization and map creation (SLAM) in unknown environment based on single sensor information, we propose a map creation method based on extended Kalman filter (EKF) for fusion of lidar and depth camera data. Primary, we convert the depth image information into pseudo-laser data, and when performing data fusion, we treat the lidar and pseudo-laser data as different measurement sources, change the input of the measurement model into multiple lidar data, and then fuse the different laser data using the measurement update step of the EKF. When performing the measurement update step, the measurement error and noise of different measurement sources may be different, and weighting factors need to be defined to weigh the contributions of different laser data to better fuse the different laser data. It can be demonstrated experimentally that the use of a single lidar can only detect obstacles parallel to the lidar, while the use of lidar fused with a depth camera can detect obstacles below the lidar plane, effectively improving the integrity and accuracy of robotic map construction.