Simon Stepputtis
Assistant Professor | Virginia Tech
I am an Assistant Professor in the Department of Mechanical Engineering at Virginia Tech, with a courtesy appointment in the Department of Computer Science, where I lead the Thinking Embodied Agents (TEA) Lab. I build robots that operate in human environments rather than around them. In the TEA Lab, we pair learned policies with symbolic reasoning so robots can acquire skills over their lifetime, adapt to the people they work with, and stay safe in contact.
Selected Work
Language Movement Primitives: Grounding Language Models in Robot Motion
- arXiv 2026
- Vision-language models set the parameters of dynamic movement primitives, turning free-form instructions into stable robot trajectories.
- IROS 2024
- Decomposing an unfamiliar object into simple convex parts lets a language model reason about where to grasp it for a given task.
- EMNLP 2023
- Tracking deception across long social-deduction dialogues, and measuring where language models fall short of human role inference.
A System for Imitation Learning of Contact-Rich Bimanual Manipulation Policies
- IROS 2022
- Learning contact-rich bimanual insertion from teleoperated demonstrations, where force and torque feedback turns out to be the critical signal.
Language-Conditioned Imitation Learning for Robot Manipulation Tasks
- NeurIPS 2020
- Pairing demonstrations with verbal descriptions yields visuomotor policies that can be redirected at runtime by new language commands.
Full publication list on Google Scholar and at the TEA Lab.