The multi_robot_distributional_rl_navigation from robustfieldautonomylab

robustfieldautonomylab / multi_robot_distributional_rl_navigation Goto Github PK

View Code? Open in Web Editor NEW

[ICRA 2024] Decentralized Multi-Robot Navigation for Autonomous Surface Vehicles with Distributional Reinforcement Learning

License: MIT License

Python 100.00%

multi_robot_distributional_rl_navigation's People

Contributors

Stargazers

Watchers

multi_robot_distributional_rl_navigation's Issues

ValueError in save_evaluation Method of trainer.py

Description:

I encountered a ValueError while running the script train_rl_agents.py. The error occurs when trying to save an evaluation using np.savez in the save_evaluation method of trainer.py. The specific error message is as follows:

Traceback (most recent call last):
  File "train_rl_agents.py", line 121, in <module>
    run_trial(args.device,param)
  File "train_rl_agents.py", line 100, in run_trial
    trainer.learn(total_timesteps=params["total_timesteps"],
  File "/home/hb/RLNav/Multi_Robot_Distributional_RL_Navigation/policy/trainer.py", line 179, in learn
    self.save_evaluation(eval_log_path)
  File "/home/hb/RLNav/Multi_Robot_Distributional_RL_Navigation/policy/trainer.py", line 371, in save_evaluation
    np.savez(
  File "<__array_function__ internals>", line 200, in savez
  File "/home/hb/anaconda3/envs/1_body/lib/python3.8/site-packages/numpy/lib/npyio.py", line 615, in savez
    _savez(file, args, kwds, False)
  File "/home/hb/anaconda3/envs/1_body/lib/python3.8/site-packages/numpy/lib/npyio.py", line 716, in _savez
    val = np.asanyarray(val)
ValueError: setting an array element with a sequence. The requested array has an inhomogeneous shape after 4 dimensions. The detected shape was (1, 1, 3, 1002) + inhomogeneous part.

Steps to Reproduce:

python train_rl_agents.py -C config/config_IQN.json -D cuda
The error occurs just after evaluation.

Expected Behavior:

The evaluation npz should be saved successfully without any errors.

Actual Behavior:

A ValueError is raised, indicating an issue with the shape of the array being saved.

Environment:

Python version: 3.8
gym version: 0.19.0
Operating System: Ubuntu 20.04.6 LTS x86_64

Additional Information:

It seems that the issue is related to the shape of the array being passed to np.savez.
Upon trying to convert values of data_to save using np.array(), I found that data_to_save['objs'] has an incorrect shape, which likely causes this issue.