How can I verify the policy?

[kimyg119]

Hello, thanks for reading my issue.

Now I'm running imitation learning myself with example you provided, and as far as I know, there will be a checkpoint after learning is done. How can I check if it is working well on isaac sim?

[xshenhan]

Hi, I've updated the eval code, a pretrained policy and some instructions in ./re3sim/eval.md. You can have a try!

[VictoriYang]

Hi ~ A class called FrankaTableMultiItemPickAndPlaceForEnv used in scripts/generate_test_cases.py doesn't exist in this code repo:
from real2sim2real.tasks.multi_item_pick_and_place import ( FrankaTableMultiItemPickAndPlaceForEnv, )

[xshenhan]

Hi. Maybe you can just remove the import like this and change the config yaml file to the existing one. I'll have a test now and will fix the inconsistency soon.

From:

from real2sim2real.tasks.multi_item_pick_and_place import (
    FrankaTableMultiItemPickAndPlaceForEnv,
)
from real2sim2real.randomize.item_pose_random import random_place_items


env.reset()
task: FrankaTableMultiItemPickAndPlaceForEnv = list(env.runner.current_tasks.values())[
    0
]

To:

from real2sim2real.randomize.item_pose_random import random_place_items


env.reset()
task = list(env.runner.current_tasks.values())[
    0
]