Currently, NoAction detection is based on the lack of action change over multiple frames. In this setup, objects in uniform or uniformly accelerated motion are mistakenly identified as inactive.
To Reproduce
Interact with the env using the same action for more than 100 steps.
Expected behavior
The detection should only consider an object inactive when there is a prolonged lack of change in the object's state, rather than just its actions.
Describe the bug
Currently, NoAction detection is based on the lack of action change over multiple frames. In this setup, objects in uniform or uniformly accelerated motion are mistakenly identified as inactive.
To Reproduce Interact with the env using the same action for more than 100 steps.
Expected behavior The detection should only consider an object inactive when there is a prolonged lack of change in the object's state, rather than just its actions.