IanYangChina / pybullet_multigoal_gym

Pybullet version of the multigoal robotics environment from OpenAI Gym
Apache License 2.0
82 stars 17 forks source link

Several minor issues #5

Closed SamNPowers closed 2 years ago

SamNPowers commented 2 years ago

Hi, thanks for the helpful repo! It's exactly what I was looking for.

I'm encountering a few issues though.

  1. The examples define observation_cam_id as an integer (e.g. 0), but https://github.com/IanYangChina/pybullet_multigoal_gym/blob/c5a90680702f17734cf1c9b1f797b8ac9846810e/pybullet_multigoal_gym/__init__.py#L116 assumes it's a list. (Based on the default value assigned at the beginning of the function it should be, e.g. [0])
  2. obs_dict as used https://github.com/IanYangChina/pybullet_multigoal_gym/blob/c5a90680702f17734cf1c9b1f797b8ac9846810e/pybullet_multigoal_gym/envs/base_envs/kuka_multi_step_base_env.py#L341 isn't defined before it is used
  3. Joints as used here https://github.com/IanYangChina/pybullet_multigoal_gym/blob/c5a90680702f17734cf1c9b1f797b8ac9846810e/pybullet_multigoal_gym/robots/kuka.py#L341 expect 2 outputs, but get_state() returns 3 (x, vx, fx).
  4. Observation isn't defined when it is used here: https://github.com/IanYangChina/pybullet_multigoal_gym/blob/c5a90680702f17734cf1c9b1f797b8ac9846810e/pybullet_multigoal_gym/envs/base_envs/kuka_multi_step_base_env.py#L347

And likely more; I stopped here as this is more errors than I expect to find. Is there a different version of the codebase I should be using? Am I missing something more obvious?

Thanks!

IanYangChina commented 2 years ago

Hi, will look into them. These are probably caused by my experimental codes. I have been doing some tinkerings and did not clean up things properly. I would suggest you use the released version, probably less or no errors. Are you using the master branch?

SamNPowers commented 2 years ago

I am on master. I'll use the latest released version instead for now (v1.2 I believe?).

Thanks!

IanYangChina commented 2 years ago

Hi, I fixed all the bugs and released a new version (v1.3). Or you can just keep using the master branch now. Thanks.