search

JonathanYang0127 / omnimimic

Cross-Embodiment Robot Learning Codebase
35 stars 2 forks source link

About The Topological Map #3

Open ChengYuqing1998 opened 7 months ago

ChengYuqing1998 commented 7 months ago

Hi jonathan, I've read the paper and been attacted by your work since I also work in the field of robotic navigation.

In the paper's appendix C "Navigation Details" part, it says that one can get a topological map of goals by converting the recording video. I searched the codes but I didn't find any funcs or scripts covering for that. I wonder whether such converting codes would be open sourced in the future since the topological map seems to be pivotal to apply goal image conditioned model, like ViNT, to navigate to long horizon goal.

Thanks!

JonathanYang0127 commented 7 months ago

Hi Cheng, the topological map code should be found in the ViNT codebase https://github.com/robodhruv/visualnav-transformer/blob/main/deployment/src/create_topomap.py