17/09/2025
Looking for an easy-to-use way to deploy your VLAs and diffusion policies on real robots (maybe even before the ICRA video deadline next week ๐)?
๐ Excited to share our latest work: ๐๐ฅ๐๐ฆ๐ฃ - ๐๐ผ๐บ๐ฝ๐น๐ถ๐ฎ๐ป๐ ๐ฅ๐ข๐ฆ2 ๐๐ผ๐ป๐๐ฟ๐ผ๐น๐น๐ฒ๐ฟ๐ ๐ณ๐ผ๐ฟ ๐๐ฒ๐ฎ๐ฟ๐ป๐ถ๐ป๐ด-๐๐ฎ๐๐ฒ๐ฑ ๐ ๐ฎ๐ป๐ถ๐ฝ๐๐น๐ฎ๐๐ถ๐ผ๐ป ๐ฃ๐ผ๐น๐ถ๐ฐ๐ถ๐ฒ๐ ๐ฎ๐ป๐ฑ ๐ง๐ฒ๐น๐ฒ๐ผ๐ฝ๐ฒ๐ฟ๐ฎ๐๐ถ๐ผ๐ป!
Why CRISP?
Most learning-based policies (diffusion, VLAs, etc.) output low-frequency or discontinuous commands, which donโt play nicely with real hardware. CRISP bridges that gap with lightweight, compliant, torque-based controllers built on hashtag control.
Features:
๐ ๐ฃ๐๐๐ต๐ผ๐ป ๐ถ๐ป๐๐ฒ๐ฟ๐ณ๐ฎ๐ฐ๐ฒ โ control your robot around without worrying about topics, spinning, etc., but still keep full ROS2 flexibility.
๐ ๐๐๐บ๐ป๐ฎ๐๐ถ๐๐บ ๐ฒ๐ป๐๐ถ๐ฟ๐ผ๐ป๐บ๐ฒ๐ป๐ โ use teleoperation to record robot and visual data in hashtag format & seamlessly deploy learning-based policies.
๐ฅ ๐๐ฒ๐บ๐ผ-๐ฟ๐ฒ๐ฎ๐ฑ๐ โ examples for single-arm & bimanual manipulation with the Franka FR3.
๐ค ๐๐ฎ๐ฟ๐๐ฒ๐๐ถ๐ฎ๐ป & ๐ท๐ผ๐ถ๐ป๐-๐๐ฝ๐ฎ๐ฐ๐ฒ ๐ฐ๐ผ๐ป๐๐ฟ๐ผ๐น๐น๐ฒ๐ฟ๐ โ built for torque control and smooth, compliant interaction.
๐ซ ๐ก๐ผ ๐ ๐ผ๐๐ฒ๐๐ ๐ผ๐ฟ ๐ต๐ฒ๐ฎ๐๐ ๐ฝ๐น๐ฎ๐ป๐ป๐ถ๐ป๐ด ๐๐๐ฎ๐ฐ๐ธ๐ required, ready to use.
Our goal: Make deploying learning-based methods on real robots as frictionless as possible, reducing the gap between data collection, simulation, and deployment.
๐ Paper: https://arxiv.org/abs/2509.06819
๐ป Code: https://github.com/utiasDSL/crisp_controllers
๐ Website: https://utiasdsl.github.io/crisp_controllers/
This work is the result of a great team effort by Daniel San Josรฉ Pro, Oliver Hausdรถrfer, Ralf Rรถmer, Maximilian Dรถsch, and Martin Schuck!
