GR1 Open Microwave Door Task#

This example demonstrates the complete workflow for the GR1 manipulation task of opening a microwave door in Isaac Lab - Arena, covering environment setup and validation, teleoperation data collection, data generation with Isaac Lab Mimic, policy post-training, and closed-loop evaluation.

../../../_images/kitchen_gr1_arena.gif

Task Overview#

Task ID: gr1_open_microwave

Task Description: The GR1T2 humanoid uses its upper body (arms and hands) to reach toward a microwave, and open the door.

Key Specifications:

Property

Value

Tags

Table-top manipulation

Skills

Reach, Open door

Embodiment

Fourier GR1T2 (54 DOF humanoid)

Interop

Isaac Lab Teleop, Isaac Lab Mimic

Scene

Kitchen environment

Objects

Microwave (articulated object)

Policy

GR00T N1.5 (vision-language foundation model)

Post-training

Imitation Learning

Dataset

Arena-GR1-Manipulation-Task

Checkpoint

GN1x-Tuned-Arena-GR1-Manipulation

Physics

PhysX (200Hz @ 4 decimation)

Closed-loop

Yes (50Hz control)

Metrics

Success rate, Door moved rate

Workflow#

This tutorial covers the pipeline between creating an environment, generating training data, fine-tuning a policy (GR00T N1.5), and evaluating the policy in closed-loop. A user can follow the whole pipeline, or can start at any intermediate step by downloading the pre-generated output of the preceding step(s), which we provide (described in the relevant step below).

Prerequisites#

Start the isaaclab docker container

./docker/run_docker.sh

We store data on Hugging Face, so you’ll need log in to Hugging Face if you haven’t already.

hf auth login

You’ll also need to create the folders for the data and models. Create the folders for the data and models with:

export DATASET_DIR=/datasets/isaaclab_arena/static_manipulation_tutorial
mkdir -p $DATASET_DIR
export MODELS_DIR=/models/isaaclab_arena/static_manipulation_tutorial
mkdir -p $MODELS_DIR

Workflow Steps#

Follow the following steps to complete the workflow: