Open wlwsjl opened 1 month ago
Hi, Thank you for your interest. You’re right. The image is mostly 5Hz, and the IMU is mostly 100Hz. I’ll update the website. Thank you for pointing that out. I think these documents might be helpful if you need more information on sensor setup. https://ntrs.nasa.gov/api/citations/20160007769/downloads/20160007769.pdf https://ntrs.nasa.gov/api/citations/20190030473/downloads/20190030473.pdf
Best,
2024년 7월 31일 (수) 오전 12:46, wlwsjl @.***>님이 작성:
Hi, thanks for the release of this interesting dataset. I downloaded two sequences, iva_ARtag and ff_return_journey_forward. And I found image frequency is around 5 hz (< 15 hz), imu frenquency is also below 250 hz. Can you check if these datasets uploaded to the website are right? Looking forward to your reply!
— Reply to this email directly, view it on GitHub https://github.com/suyoung1222/Astrobee-Dataset-Tool/issues/1, or unsubscribe https://github.com/notifications/unsubscribe-auth/A36URVD7LY7ZJWT2FQPMVSLZO6YPFAVCNFSM6AAAAABLWU65A2VHI2DSMVQWIX3LMV43ASLTON2WKOZSGQZTQMJVGMYTMOI . You are receiving this because you are subscribed to this thread.Message ID: @.***>
Thanks for your reply and shared docs. I find another issue with zip file. For iva_kibo_rot sequence, 1652460792706759936 exists in gray.txt, but does not exist in gray.zip. Please check all sequences.
https://github.com/nasa/astrobee/issues/790 Btw, as authors of AstroLoc are cleaning their code, we are unable to use your baseline easily to reproduce localization results. Can you release other open-sourced VIO framework you tested, like ORB-SLAM3? I believe this would be more convenient and transparent for researchers to use novel Astrobee ISS dataset. And it's also an effective way to ensure all sequences of dataset are good to use.
Best regards,
Hi, thanks for the update about sensor frequency on the website. Are results in the table III of paper also obtained with image at 5 Hz, and IMU at 100 Hz or even lower than 100 Hz?
Hi, We fixed the text file issue. Images are processed at 5 Hz. The AstroLoc system processes IMU data at 62.5 Hz, as mentioned in the AstroLoc paper. In the benchmark, only AstroLoc uses IMU, while the others are monocular.
2024년 8월 8일 (목) 오전 2:18, wlwsjl @.***>님이 작성:
Hi, thanks for the update about sensor frequency on the website. Are results in the table III of paper also obtained with image at 5 Hz, and IMU at 100 Hz or even lower than 100 Hz?
— Reply to this email directly, view it on GitHub https://github.com/suyoung1222/Astrobee-Dataset-Tool/issues/1#issuecomment-2273954235, or unsubscribe https://github.com/notifications/unsubscribe-auth/A36URVEV6NMZCU3YZYJN4UTZQJJIHAVCNFSM6AAAAABLWU65A2VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDENZTHE2TIMRTGU . You are receiving this because you commented.Message ID: @.***>
Hi, I'm curious why not benchmark with other VIO algorithms. As mentioned in the abstract, this dataset is targeted at visual-inertial localization. Did you also evaluate AstroLoc with Sim(3) mode? https://github.com/uzh-rpg/rpg_trajectory_evaluation
Hi, thanks for pointing that out. Since it doesn't have a gravity vector, it can be difficult to utilize VIO methods. However, we also believe that researching the use of IMU in a zero-g environment can be interesting future work. Yes, we used the scaled SIM3 mode with this repo: https://github.com/MichaelGrupp/evo/wiki/Metrics
2024년 8월 8일 (목) 오후 9:38, wlwsjl @.***>님이 작성:
Hi, I'm curious why not benchmark with other VIO algorithms. As mentioned in the abstract, this dataset is targeted at visual-inertial localization. Did you also evaluate AstroLoc with Sim(3) mode? https://github.com/uzh-rpg/rpg_trajectory_evaluation
— Reply to this email directly, view it on GitHub https://github.com/suyoung1222/Astrobee-Dataset-Tool/issues/1#issuecomment-2275719975, or unsubscribe https://github.com/notifications/unsubscribe-auth/A36URVBUENOLMVZJDGHB72LZQNRGBAVCNFSM6AAAAABLWU65A2VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDENZVG4YTSOJXGU . You are receiving this because you commented.Message ID: @.***>
Thank you very much for your answer.
Hi, thanks for the release of this interesting dataset. I downloaded two sequences, iva_ARtag and ff_return_journey_forward. And I found image frequency is around 5 hz (< 15 hz), imu frenquency is also below 250 hz. Can you check if these datasets uploaded to the website are right? Looking forward to your reply!