I need help interpreting some data. Here is the leaderboard for the performance of different odometry/SLAM approaches, benchmarked against the KITTI Vision dataset: The top approach, V-LOAM, uses lidar. The fourth best approach, SOFT2, uses stereo camera images, and no lidar data. The rotation error is the same for SOFT2 and V-LOAM, and the translation error for SOFT2 is 0.05% higher. On the face of it, this would seem to imply that the current best stereo camera-based odometry/SLAM method is pretty close in accuracy to the current beat lidar-based method. Am I wrong? Is is a 0.05% difference actually huge? Am I otherwise misunderstanding or misinterpreting what’s going on here? Any help anyone can provide would be much appreciated. Thanks.