[Question] BC training obs key request

[longann890621]

Question

Hi, I have a question：
I'm currently doing imitation learning training, but the training effect is not good (I can't deduce the correct action in the scene).
I had already learned that my HDF5 file must have OBS and Action, as shown in the image below.

And my configuration file bc_rnn_image_84.json needs to have the corresponding OBS key, as shown in the image below.

I'd like to know if the OBS format specifies what kind of keys should be included in order to achieve good imitation learning training? Or, approximately how many lines of correct action demonstration data are needed for good imitation learning and correct inference in the scene?

[longann890621]

Additional notes:

1. My task is to use the TM5 robotic arm to clamp the block on the table into the box next to it.
2. My current actions data in HDF5 is: [△x, △y, △z, △roll, △pitch, △yaw,gripper].
   I wonder know if my format is correct?
   Or should it be : [normalize△x, normalize△y, normalize△z, normalize△roll, normalize△pitch, normalize△yaw, normalize gripper]

[RandomOakForest]

Thank you for posting this. I'll move this post to our Discussions section for follow up.