The modified Panda-gym benchmark for evaluating skill-aware RL algorithms.
git clone https://github.com/uoe-agents/Skill-aware-Panda-gym.git
pip install -e .import skill_aware_panda_gym
import gymnasium as gym
# almost same with panda-gym
# but show skill by info
env = gym.make('PandaSkillAware-v3', render_mode="rgb_array", friction=1.0, mass=1.0)
observation, info = env.reset()
states = []
for _ in range(1000):
action = env.action_space.sample() # random action
observation, reward, terminated, truncated, info = env.step(action)
# info['state'] ∈ {'push', 'roll', 'pick','down'}
states.append(info['state'])
if terminated or truncated:
observation, info = env.reset()
if 'pick' in states: print('pick skill')
elif 'roll' in states: print('roll skill')
else: print('push skill')
states.clear()
env.close()