By Andrew Raffi Ansell & Mark Garo Ansell, Sidekick Robotics, Inc.
Enter here→At Sidekick Robotics, we have always been interested in bringing lessons from Large Language Models (Gen AI) into Vision Language Action Models (Physical AI).
Our preliminary internal results suggest that prompting a VLA changes model behavior, much like in-context learning with LLMs. It is a promising signal, and we want to explore it further.
Before publishing our findings, we want to learn alongside the robotics community.
01What is 💬 PromptVLA
💬 PromptVLA is an open challenge to map the behaviors and limits of prompted VLAs.
02Who can participate
Teams with access to robots running Vision Language Action foundation models
Ability to modify the prompt being passed to the VLA
03What you get
Early access to the full results compiled by our team at Sidekick Robotics
Recognition in the writeup and consideration for co-authorship
04How to participate
Enter here
Sign up to take part in the 💬 PromptVLA challenge
Please share widely.
Happy prompting,
Andrew Raffi Ansell & Mark Garo Ansell
Sidekick Robotics, Inc.