SARA-RT: Scaling up Robotics Transformers with Self-Adaptive Robust Attention
This addresses the problem of deploying large-scale robotic policies efficiently for robotics applications, though it appears incremental as it builds on existing Transformer methods.
The paper tackles the challenge of scaling up Robotics Transformers for on-robot deployment by introducing SARA-RT, which converts quadratic-time Transformer policies into efficient linear-attention versions while maintaining high quality, demonstrated by speeding up RT-2 and Point Cloud Transformer models.
We present Self-Adaptive Robust Attention for Robotics Transformers (SARA-RT): a new paradigm for addressing the emerging challenge of scaling up Robotics Transformers (RT) for on-robot deployment. SARA-RT relies on the new method of fine-tuning proposed by us, called up-training. It converts pre-trained or already fine-tuned Transformer-based robotic policies of quadratic time complexity (including massive billion-parameter vision-language-action models or VLAs), into their efficient linear-attention counterparts maintaining high quality. We demonstrate the effectiveness of SARA-RT by speeding up: (a) the class of recently introduced RT-2 models, the first VLA robotic policies pre-trained on internet-scale data, as well as (b) Point Cloud Transformer (PCT) robotic policies operating on large point clouds. We complement our results with the rigorous mathematical analysis providing deeper insight into the phenomenon of SARA.