robodhruv / visualnav-transformer

Official code and checkpoint release for mobile robot foundation models: GNM, ViNT, and NoMaD.
http://general-navigation-models.github.io
MIT License
425 stars 56 forks source link

What do the four values of choosen_waypoint represent? #11

Closed Hwh0865 closed 6 months ago

Hwh0865 commented 6 months ago

Very nice work! When deploying, if choosen_waypoint=[ 0.15498, -0.00129, 0.9395, -0.3425 ], the robot needs linear and angular velocities to be driven, how can I calculate the linear and angular velocities from the four data, and what do these four data mean? Thank you!

wmh02240 commented 5 months ago

Hi, have you deployed this model on your robot? How is the effect? I deployed the NoMad model, but the effect is not good, collisions often occur。 Looking forward to your reply。 @Hwh0865

Hwh0865 commented 5 months ago

Good afternoon! Glad to hear of your successful deployment. I've just successfully deployed ViNT to my robot and am having the same collision problem. The current speculation is that the problem is two-fold, on one hand it may be the number of pictures in the topology map (only nine), and on the second hand the vint model has not been exposed to this unknown environment and is not able to realize the navigation function, it may need to re-collect the data from the current environment and then fine-tune it. Looking forward to your reply, thanks!

wmh02240 commented 5 months ago

Thanks for your reply, I have collected the date from my office and fine-tune the NoMad model , the same problem of collosion also occur. Have you resolve this problem? Looking forward to your reply, Thanks. @Hwh0865

Hwh0865 commented 5 months ago

哥你好,你是怎么微调的,是采完数据然后训练得到的新模型,就算微调结束了吗,需要有什么注意的么 @wmh02240

robodhruv commented 5 months ago

@Hwh0865 : In case you are still looking for an answer, pd_controller.py handles the low-level control from waypoints to velocities. https://github.com/robodhruv/visualnav-transformer/blob/main/deployment/src/pd_controller.py

@wmh02240 : Discussing your issue here https://github.com/robodhruv/visualnav-transformer/issues/13