Kirill Utyashev

I’m seeking PhD opportunities in robot learning for Fall 2027!

Hi! I’m a fourth-year undergraduate in Computer Science at the University of Toronto. I’m fortunate to work with Prof. Florian Shkurti in the Robot Vision and Learning Lab, Prof. Peter Marbach on multi-agent reinforcement learning, and Prof. Avi Goldfarb on AI and user behavior.

My primary interests lie in robot learning, reinforcement learning, and manipulation, with the goal of developing robots that operate safely and reliably in the real world. At RVL, I investigate the visual robustness of vision-language-action models and how force feedback can improve their reliability in contact-rich manipulation.

I’m always happy to chat about research or shared interests! Feel free to reach out at kirill.utyashev@mail.utoronto.ca.

Outside research, I enjoy Brazilian jiu-jitsu (blue belt), soccer, literature, and making beats.

Two robotic arms with grippers above a tabletop manipulation setup

Force-Aware Manipulation

Ongoing

We investigate how force feedback can improve VLA reliability in contact-rich manipulation, using fast controllers to correct pretrained policy actions.

[Poster]

Policy-Aware Curricula from Sparse-View Robotics Data

Ongoing

We investigate improving VLA visual robustness through generated scene variations tailored to the policy’s current abilities.

Oopsie Data

Ongoing

A multi-lab effort to build a dataset of real-world robot manipulation failures and suboptimal behavior. I contributed over 1,000 episodes totaling 11 hours of robot data.

[Website]