Simon Stepputtis

About

Simon Stepputtis is a robotics and artificial intelligence researcher whose work sits at the intersection of imitation learning, human-robot interaction, and robot manipulation. His research addresses one of the central challenges in modern robotics: enabling robots to learn complex, contact-rich behaviors from human demonstrations rather than explicit programming. Stepputtis's most influential contribution — a system for learning continuous human-robot interactions from human-human demonstrations (2017, 69 citations) — established a data-driven framework using motion capture to transfer interaction dynamics between humans and robots. This foundational work seeded a productive line of research spanning one-shot handover learning, probabilistic multimodal modeling, and bimanual manipulation, each advancing robots' ability to generalize across varied conditions. His 2020 paper on language-conditioned imitation learning introduced natural language as a communication channel between human experts and robotic learners, a prescient direction given subsequent developments in large language models. More recently, Stepputtis has expanded into zero-shot grasping with geometric reasoning and scene graph generation, reflecting a broader vision of robots that understand and act within complex visual environments. His comparative analysis of imitation learning algorithms also demonstrates a rigorous commitment to benchmarking that benefits the wider robotics community. Collectively, his publications have garnered over 225 citations, marking him as an emerging and increasingly versatile voice in intelligent robotics research.

Research Focus

Key Achievements

Top Papers

1. 1
   A system for learning continuous human-robot interactions from human-human demonstrations

   69 citations · 2017
2. 2
   One-shot learning of human–robot handovers with triadic interaction meshes

   26 citations · 2018
3. 3
   Probabilistic Multimodal Modeling for Human-Robot Interaction Tasks

   24 citations · 2019
4. 4
   Language-Conditioned Imitation Learning for Robot Manipulation Tasks

   21 citations · 2020
5. 5
   A System for Imitation Learning of Contact-Rich Bimanual Manipulation Policies

   18 citations · 2022
6. 6
   A Comparison of Imitation Learning Algorithms for Bimanual Manipulation

   17 citations · 2024
7. 7
   HiKER-SGG: Hierarchical Knowledge Enhanced Robust Scene Graph Generation

   15 citations · 2024
8. 8
   ShapeGrasp: Zero-Shot Task-Oriented Grasping with Large Language Models through Geometric Decomposition

   14 citations · 2024
9. 9
   Extrinsic Dexterity Through Active Slip Control Using Deep Predictive Models

   12 citations · 2018
10. 10
    Learning human-robot interactions from human-human demonstrations (with applications in Lego rocket assembly)

    11 citations · 2016

Key Collaborators

Contact & Links