This T-RO paper will be presented at #ICRA2025 develops a Contact-Implicit MPC method for controlling #robots making/breaking contact. The bilevel planner uses LCPs to reason about contact timing/forces online, generating new contact modes in real-time.

ieeexplore.ieee.org/document/103...

Did a fun lecture this week on data-drive control with hints of behavior cloning and diffusion policies, but all grounded in classical linear systems theory. Most of the ideas are drawn from the behavioral control theory literature: youtu.be/fnajY4Ip13w?...

Had a great visit to MIT for the robotics seminar a couple of weeks ago: www.youtube.com/watch?v=eSle...

Just made first contact with our satellite PROMETHEUS-1 in collaboration with @cmurexlab.bsky.social @istecnico.bsky.social @uminho.bsky.social @cmurobotics.bsky.social @scsatcmu.bsky.social

See where our satellite is here: tinygs.com/satellite/PR...

Cool work by my student @fogmb.bsky.social trying to scale joint robot design and control optimization to realistic problems. These are super challenging optimization problems that are often handled separately by expert designers. We're trying to automate it.

🚀 Co-Optimizing Robots for Performance 🚀

🌟 CAPO: A scalable optimization method:
✅ Jointly optimizes actuators & control traj.
✅ Scales to complex systems
✅ Finds robust, efficient designs

🏋️ Tested on:
🚁 Multirotors 🛰️ Space structures 🧥 Cloth 🐟 Soft robots

📄 rexlab.ri.cmu.edu/papers/CAPO_...

End of an era as Kevin Tracy defends his PhD. Kevin has been with the lab since the very beginning at Stanford, through the move to CMU. He's done amazing optimization-based control work with applications spanning aerospace and robotics and it has been an absolute pleasure working with him.