Welcome to the 2023 AI2-THOR Rearrangement Challenge hosted at the CVPR22 Embodied-AI Workshop. The goal of this challenge is to build a model/agent that can move objects in a room to restore them to a given initial configuration. Please follow the instructions below to get started.

What’s New in the 2023 Challenge?

Our 2023 AI2-THOR Rearrangement Challenge has several upgrades distinguishing it from the 2022 version:

• New AI2-THOR version: Upgrade from version 5.0.0, bringing performance improvements and bug fixes.
• New dataset: Released a new rearrangement dataset with a balanced mix of easy and hard episodes.
• Improved object-opening logic: All openable objects have toggled states, improving interaction realism.
• Misc. improvements: Fixed various minor bugs and performance issues from the previous challenge.

Getting Started with Installation

To begin, clone the repository locally:

bash
git clone git@github.com:allenaiai2thor-rearrangement.git

Local Installation

First create a Python virtual environment and then install requirements:

bash
pip install -r requirements.txt

Docker Installation

This assumes some familiarity with Docker. If you’re new to Docker, we recommend reading through this tutorial.

After installing nvidia-docker, build the Docker image:

bash
DOCKER_BUILDKIT=1 docker build -t rearrangement:latest .

Run the Docker container:

bash
docker run --gpus all -it rearrangement:latest

Understanding the Rearrangement Task

Think of the Rearrangement Challenge like a game of Tetris, but instead of fitting pieces together, your challenge is to properly arrange a room’s furniture! Your agent first observes the room in a perfect layout during the walkthrough phase. Then, it encounters a chaotic unshuffle phase where a few pieces are moved. The objective? Reset everything back to the ideal structure.

Submitting to the Leaderboard

Participants can track their progress on the AI2 Leaderboard. Ensure your submission is uniquely different from baseline models and prior entries. All metrics can be extracted from your agent’s actions across tasks stored in the combined.pkl.gz dataset.

Troubleshooting

If you run into any issues, consider these troubleshooting steps:

• Check your installation paths and dependencies.
• Ensure your Docker image is correctly configured and running.
• Verify the integrity of datasets being used.

For more insights, updates, or to collaborate on AI development projects, stay connected with fxis.ai.

Conclusion

At fxis.ai, we believe that such advancements are crucial for the future of AI, as they enable more comprehensive and effective solutions. Our team is continually exploring new methodologies to push the envelope in artificial intelligence, ensuring that our clients benefit from the latest technological innovations.