search

wzzheng / OccWorld

[ECCV 2024] 3D World Model for Autonomous Driving
https://wzzheng.net/OccWorld/
Apache License 2.0
382 stars 25 forks source link

Quesion about the Pose_mode #14

Closed yuyang-cloud closed 7 months ago

yuyang-cloud commented 10 months ago

Excellent Work and released code! I went through the code and met an unclear question in dataset.py at here: gt_modes.append(self.nusc_infos[scene_name][idx+i]['pose_mode'])

The pose_mode seems to be a three-dimensional vector. What does it mean? Could authors give an explanation? Thanks in advance!

SeaBird-Go commented 10 months ago

I also want to ask this question, what are the meanings of rel_poses and gt_mode in the pickle file? Since we don't know how to create these pickle files.

synsin0 commented 9 months ago

same question.

suixin1424 commented 8 months ago

same question.

yuyang-cloud commented 7 months ago

Hi! Is there anyone who can answer this question? How to generate pkl file in this issue? What are the meanings of rel_poses and gt_mode? @wzzheng Many thanks in advance!

chen-wl20 commented 7 months ago

Thank you very much for your attention to the repository for this job. I apologize for the late response. Regarding the part on motion planning, we borrowed from the work of vad in the pkl file and evaluation metrics, so you can refer to vad on how to generate pkl files. As for pose_mode, I borrowed the gt_ego_fut_cmd from the pkl file of vad. Therefore, based on my understanding, pose predicts three possible future displacements (I believe they are left turn, straight, and right turn), and cmd is used to select the actual situation. Then, a comparison is made with the actual displacement to obtain the supervision signal. Thank you again for your attention.

yuyang-cloud commented 7 months ago

Thanks for your reply!