Tewodros W. Ayalew

Email: tewodrosayalew [at] uchicago [dot] edu

I am a PhD student in Computer Science at the University of Chicago, advised by Dr. Matthew Walter at the Robot Intelligence through Perception Laboratory (RIPL) at the Toyota Technological Institute at Chicago (TTIC).

My research focuses on world models and latent actions learned from diverse, unlabeled video, with reward learning from visual observations as a complementary thread. In CLAW, I jointly learn a diffusion-based world model and continuous latent actions end-to-end from action-free video; the resulting model supports imitation from observation and goal-directed planning, and I evaluated it on a real UR5 with pretraining on human video (e.g., 43% RoboDesk upright-block planning vs. 5% AdaWorld). In PROGRESSOR (ICCV 2025), I learn a self-supervised visual progress reward with online refinement for robot RL without manual reward design. I also co-authored a CoRL 2026 Oral on what robot manipulation benchmarks actually measure. I am interested in scaling these directions beyond curated datasets toward in-the-wild visual experience.

Prior to my PhD, I earned an M.Sc. in Computer Science from the University of Saskatchewan, where I worked with Dr. Ian Stavness on unsupervised computer vision.

I am seeking Summer 2027 research internships in robot learning and world models.