Open HeXu1 opened 3 years ago
@HeXu1 Sorry, I cannot provide the test data for now, but I will provide it if I have some spare time. You're right. There was something wrong in the lidar to camera extrinsic in my yaml file (I've fixed it), but it's not used in the code (only translation is used) so no influence I guess.
"vins_world" to "vins_body_ros" transform is the pose of the LiDAR estimated by the IMU pose estimate computed in VINS (in visual_odometry/visual_estimator/utility/visualization.cpp). Rotation is not used because, in the function get_depth() in /visual_odometry/visual_feature/feature_tracker.h, only "translation" to camera frame is required to compare the depths from the LiDAR and the point features with the x-axis forward and z-axis upward like the LiDAR frame.
Hope my explanation helps you solve any issues you are going through.
If not, could you tell me what's not working out for you?
Thank you.
@epicjung ok ,I will try it with myself data.thanks!
can you upload some test data about your config, thanks. And did " lidar to camera extrinsic" rotation is opposite?