Stone Tao

Make a proper test for trajectory replay tool

5

Currently we just run the replay tool for coverage purposes. Should add tests to check the actual contents of the saved trajectories/videos.

Previously

Likely to be fixed in Sapien instead, but kept as an issue here for visibility. Originally from #130 The current fix is to make sure all cameras are using the...

Using gymnasiums vector env and context="forkserver" (other contexts do not work for me, I don't entirely know why), the code runs fine and can be verified by running the demo_vec_env.py...

When running all tests, sometimes I get ``` FAILED tests/test_wrappers.py::test_recordepisode_wrapper[state_dict-PickSingleEGAD-v0] - RuntimeError: Renderer is not created. Renderer creation is required before any other operation. ``` When testing just the test_wrappers...

The pytest that shows this is ``` @pytest.mark.parametrize("env_id", ENV_IDS) def test_states(env_id): env: BaseEnv = gym.make(env_id) obs, _ = env.reset(seed=1000) for _ in range(5): env.step(env.action_space.sample()) state = env.get_state() obs = env.get_obs()...

vulkan: No DRI3 support detected - required for presentation

2

I'm getting this message when I run the example code in the readme that generates videos. It seems in headless=False mode it can't open the viewer and the env.render function...

Getting environment and setting environment states

1

Was wondering if it is possible to get some environment state and then set the environment back to the same exact state later? I tried using `envs.gym.set_actor_root_state_tensor_indexed` and `envs.gym.set_dof_state_tensor_indexed` among...