google-research/robotics_transformer
RT-1 is a transformer model that uses image tokenization and learned token compression for end-to-end robotic control.
Not currently ranked — collecting fresh signals.
star history
Robotics Transformer (RT-1) is a Google Research project that implements a transformer architecture for controlling robots. It uses a FiLM EfficientNet-based image tokenizer to process visual inputs, a token learner for compressing input tokens, and a transformer decoder for producing robotic actions. The repository includes pre-trained checkpoints for multiple robot control scenarios including multi-robot and sim-to-real deployment.
Frequently asked
- What is google-research/robotics_transformer?
- RT-1 is a transformer model that uses image tokenization and learned token compression for end-to-end robotic control.
- Is robotics_transformer open source?
- Yes — google-research/robotics_transformer is open source, released under the Apache-2.0 license.
- What language is robotics_transformer written in?
- google-research/robotics_transformer is primarily written in Python.
- How popular is robotics_transformer?
- google-research/robotics_transformer has 1.7k stars on GitHub.
- Where can I find robotics_transformer?
- google-research/robotics_transformer is on GitHub at https://github.com/google-research/robotics_transformer.