I'm trying to train RL-agent to play Car Racing environment with OpenAI Gym and been using following code:
import gym
from stable_baselines3 import PPO
from stable_baselines3.common.vec_env import DummyVecEnv
from stable_baselines3.common.evaluation import evaluate_policy
import os
environment_name = 'CarRacing-v2'
env = gym.make(environment_name, render_mode="human")
env.action_space
env.observation_space
episodes = 1
for episode in range(1, episodes+1):
state = env.reset()
done = False
score = 0
while not done:
action = env.action_space.sample()
n_state, reward, done, info, _ = env.step(action)
score+=reward
print(f'Episode:{episode} Score:{score}')
env.close()
TRAIN MODEL:
env = gym.make(environment_name)
env = DummyVecEnv([lambda: env])
model = PPO("CnnPolicy", env, verbose=1)
model.learn(total_timesteps=1) # Just to show how a model is trained
# later we will train model over millions of timesteps
Everything works fine (except the upmost code block takes forever to run because the episode only ends after the car has driven off the map and the speed is very slow..) but the last row "model.learn(total_timesteps=1)" procudes the following error:
Box([-1. 0. 0.], 1.0, (3,), float32)
Box(0, 255, (96, 96, 3), uint8)
/Library/Frameworks/Python.framework/Versions/3.10/lib/python3.10/site-packages/tqdm/auto.py:22: TqdmWarning: IProgress not found. Please update jupyter and ipywidgets. See https://ipywidgets.readthedocs.io/en/stable/user_install.html
from .autonotebook import tqdm as notebook_tqdm
2022-11-18 15:15:08.327 Python[94428:771948] ApplePersistenceIgnoreState: Existing state will not be touched. New state will be written to /var/folders/6l/gfqkwfbd7rs176sshdhfz5f80000gn/T/org.python.python.savedState
Episode:1 Score:-866.6780141845032
Using cpu device
Wrapping the env in a VecTransposeImage.
Output exceeds the size limit. Open the full output data in a text editor
---------------------------------------------------------------------------
ValueError Traceback (most recent call last)
Cell In [9], line 1
----> 1 model.learn(total_timesteps=1)
File /Library/Frameworks/Python.framework/Versions/3.10/lib/python3.10/site-packages/stable_baselines3/ppo/ppo.py:317, in PPO.learn(self, total_timesteps, callback, log_interval, eval_env, eval_freq, n_eval_episodes, tb_log_name, eval_log_path, reset_num_timesteps, progress_bar)
303 def learn(
304 self: PPOSelf,
305 total_timesteps: int,
(...)
314 progress_bar: bool = False,
315 ) -> PPOSelf:
--> 317 return super().learn(
318 total_timesteps=total_timesteps,
319 callback=callback,
320 log_interval=log_interval,
321 eval_env=eval_env,
322 eval_freq=eval_freq,
323 n_eval_episodes=n_eval_episodes,
324 tb_log_name=tb_log_name,
325 eval_log_path=eval_log_path,
326 reset_num_timesteps=reset_num_timesteps,
327 progress_bar=progress_bar,
328 )
...
---> 94 self.buf_obs[key][env_idx] = obs
95 else:
96 self.buf_obs[key][env_idx] = obs[key]
ValueError: setting an array element with a sequence. The requested array has an inhomogeneous shape after 1 dimensions. The detected shape was (2,) + inhomogeneous part.
Why does this error happen and how can I get around it?
