google-research/robotics_transformer

Health v2 · maintenance only

Flags: no_releases archived

Adoption not part of the score

1743 stars · 205 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

A Python library from Google Research implementing RT-1 (Robotics Transformer), a transformer model for end-to-end robotic control with an image tokenizer backbone and token learner compression. It includes testing utilities and pretrained TensorFlow SavedModel checkpoints for several trained variants.

Use cases

• run RT-1 robotics transformer inference
• load pretrained RT-1 checkpoints for robot control
• train a transformer for end-to-end robotic manipulation
• tokenize camera images and actions for robot learning
• experiment with token learner architectures for robotics

When to choose

• you want to use or extend the RT-1 robotics transformer model
• you need pretrained checkpoints for robotic control research
• you're reproducing the RT-1 paper in TensorFlow

When to avoid

• you need a production robotics stack or hardware integration
• you prefer PyTorch or JAX rather than TensorFlow
• you need actively developed features or official support

Facets

library · maturity maintenance

machine-learning deep-learning transformers robotics robotics machine-learning deep-learning python rt-1 robotic-control tensorflow research-code google-research linux

Member repositories

For agents

markdown · JSON · MCP: product_card(name="google-research/robotics_transformer")