search

MAPS-Lab / OdomBeyondVision

MIT License
51 stars 5 forks source link

Ground Truth #5

Open stephen9999999 opened 1 year ago

stephen9999999 commented 1 year ago

Hi, thanks for your great works! your paper mentions that the ground-truth of handheld sequences is provided by A-LOAM. So we need to run A-LOAM for each sequence to get the ground-truth? Is the UGV and UAV ground-truth from topic "/odom" and "/vrpn_client_node/X500_pglite/pose" respetively? By the way, could you provide the ground-truth files for the dataset? and some toolkit to convert the extracted ground-truth to TUM format? thank you.

li-peize commented 1 year ago

Hi there, Thanks for your interest in our dataset.

So we need to run A-LOAM for each sequence to get the ground-truth?

We uploaded ground truth for handheld in the handheld/gt folder in the Google Cloud link. It should be the same as if you run ALOAM yourself.

Is the UGV and UAV ground-truth from topic "/odom" and "/vrpn_client_node/X500_pglite/pose" respetively?

Yes, exactly.

By the way, could you provide the ground-truth files for the dataset? and some toolkit to convert the extracted ground-truth to TUM format?

Release of extracted files rather than rosbags is in the plan. It can be expected before the IROS2022 conference.

stephen9999999 commented 1 year ago

Thanks for your reply. I still have some questions. Is the depth aligned with the RGB image? If not, how can I transform the depth to the RGB image coordinate? By the way, any update about the calibration files?(such as the intrinsic params of RGB camera, the extrinsic of sensors)