Evaluation Code for R3M

The codebase contains the evaluation codebase from the paper R3M: A Universal Visual Representation for Robot Manipulation.

It trains policies from pixels with behavior cloning using pre-collected demonstrations, evaluating the policies in the environment at regular intervals. It allows for selecting different visual representations to use during imitation.

Environment Installation

The first step to running the code involves installing the evaluation environments.

For metaworld environments, install the environments by cloning this fork of the metaworld repo and installing via pip install -e .

In order to install the Franka Kitchen and Adroit environments, first install the mjrl repo using instructions here.

Then, install the RoboHive repo as described in this tag.

https://github.com/vikashplus/robohive/releases/tag/v0.0.5

Installing R3M

To use the R3M model, simply follow the installation process in the parent directory here.

Downloading Demonstration Data

All demonstrations are located here. Then change the path here to point to where the demonstration data is located. Make sure the data is saved with the same folder structure as on the google drive, e.g. <PATH TO DEMOS>/final_paths_multiview_meta_200/<CAMERA>/<TASK>.pickle.

Install and Run Eval Code

If the above was all done correctly, you should be able to simply run pip install -e . in this directory.

Verifying Correct Installation

While running all experiments can be time consuming, a simple check to make sure things are behaving as expected is to download the demos for the kitchen sliding door task, and run:

python hydra_launcher.py hydra/launcher=local hydra/output=local env="kitchen_sdoor_open-v3" camera="left_cap2" pixel_based=true embedding=resnet50 num_demos=5 env_kwargs.load_path=r3m bc_kwargs.finetune=false proprio=9 job_name=r3m_repro seed=125

and

python hydra_launcher.py hydra/launcher=local hydra/output=local env="kitchen_sdoor_open-v3" camera="left_cap2" pixel_based=true embedding=resnet50 num_demos=5 env_kwargs.load_path=clip bc_kwargs.finetune=false proprio=9 job_name=r3m_repro seed=125

You should see R3M get ~60% success on the first eval, while CLIP will get ~30%.

Commands for All Experiments

For running kitchen environments run:

python hydra_launcher.py --multirun hydra/launcher=local hydra/output=local env=["kitchen_knob1_on-v3","kitchen_light_on-v3","kitchen_sdoor_open-v3","kitchen_ldoor_open-v3","kitchen_micro_open-v3"] camera=["left_cap2","right_cap2"] pixel_based=true embedding=resnet50 num_demos=25 env_kwargs.load_path=r3m bc_kwargs.finetune=false proprio=9 job_name=try_r3m

For running metaworld environments run:

python hydra_launcher.py --multirun hydra/launcher=local hydra/output=local env=["assembly-v2-goal-observable","bin-picking-v2-goal-observable","button-press-topdown-v2-goal-observable","drawer-open-v2-goal-observable","hammer-v2-goal-observable"] camera=["left_cap2","right_cap2","top_cap2"] pixel_based=true embedding=resnet50 num_demos=25 env_kwargs.load_path=r3m bc_kwargs.finetune=false proprio=4 job_name=try_r3m

For running the Adroit pen task:

python hydra_launcher.py --multirun hydra/launcher=local hydra/output=local env=pen-v0 camera=["view_1","top","view_4"] pixel_based=true embedding=resnet50 num_demos=25 env_kwargs.load_path=r3m bc_kwargs.finetune=false proprio=24 job_name=try_r3m

For running the Adroit relocate task:

python hydra_launcher.py --multirun hydra/launcher=local hydra/output=local env=relocate-v0 camera=["view_1","top","view_4"] pixel_based=true embedding=resnet50 num_demos=25 env_kwargs.load_path=r3m bc_kwargs.finetune=false proprio=30 job_name=try_r3m