Simon Stepputtis
Postdoctoral Fellow | Carnegie Mellon Univeristy | Robotics Institute
I am on the faculty market for fall 2025!
[]
My work focuses on developing adaptive robotic systems that efficiently learn, autonomously adapt, and operate safely in human-centric environments by leveraging neuro-symbolic methods, transforming capabilities in in-home assistance, healthcare, and manufacturing.
Paper: ShapeGrasp: Zero-Shot Task-Oriented Grasping with Large Language Models through Geometric Decomposition
[] A novel method enables robots to intuitively grasp unfamiliar objects by decomposing their shapes and utilizing large language models, achieving high success rates in experimental trials. Paper on arXivPaper: A Comparison of Imitation Learning Algorithms for Bimanual Manipulation
[] Explore how different imitation learning algorithms tackle complex industrial tasks, revealing key strengths and weaknesses in precision, efficiency, and adaptability. Paper on arXivUniversity of Washington: Invited Talk
[] I am excited to give a talk at the University of Washington about Neuro-Symbolic Robot Intelligence!Multiple ICRA Workshop Papers!
[] I will be at ICRA 2024 to present some of our most recent work. Check out the Publications!.Multiple New Papers (NeurIPS, EMNLP, CoLLAs, AURO, CVPR)
[] I updated the website with multiple new papers, including EMNLP 2023, NeurIPS 2023, and CVPR 2024, CoLLAs 2024 and the Autonomous Robots Journal.Paper: Sigma: Siamese Mamba Network for Multi-Modal Semantic Segmentation
[] Introducing Sigma, a groundbreaking Siamese Mamba network that revolutionizes AI scene understanding by combining thermal, depth, and RGB data for more accurate predictions in challenging environments. Paper on arXivPaper: Sample-Efficient Learning of Novel Visual Concepts
[] Sample-efficient extraction of novel objects, affordances, and attributes from images using symbolic domain knowledge, which will be presented at CoLLAs 2023Paper: Introspective Action Advising for Interpretable Transfer Learning
[] We propose an alternative approach to transfer learning between tasks based on action advising, which will be presented at CoLLAs 2023!RSS 2023: Articulate Robots Workshop
[] I am organizing a workshop at RSS 2023 in Daegu, Republic of Korea on Articulate Robots: Utilizing Language for Robot LearningPaper: Explainable Action Advising for Multi-Agent Reinforcement Learning
[] Our new paper will be presented at ICRA 2023 in London, England! Paper on arXivPaper: Modularity through Attention: Efficient Training and Transfer of Language-Conditioned Policies for Robot Manipulation
[] Our new paper with our collaborators at Intel will be presented at CORL 2022 in Auckland, New Zeland! Paper on OpenReviewPaper: Concept Learning for Interpretable Multi-Agent Reinforcement Learning
[] Interpretable concept learning for multi-agent robot systems: CORL 2022 in Auckland, New Zeland! Paper on OpenReviewPaper: A System for Imitation Learning of Contact-Rich Bimanual Manipulation Policies
[] Our paper in collaboration with Intrinsic got accepted to IROS 2022, which we will be presenting in Kyoto, Japan. View full paperIROS 2022: TOM4HAT Workshop
[] I organized a workshop at IROS 2022 in Kyoto, Japan on Theory of MindWorkshop: RSS Pioneers
[] I was accepted to the RSS Pioneers Workshop 2022 with my work on Language-Conditioned Human-Agent Teaming.Postdoctoral Fellow at Carnegie Mellon University
[] I started as a postdoctoral fellow at Carnegie Mellon University (CMU) with Prof. Katia Sycara.Graduation: Ph.D. in Computer Science
[] I completed my Ph.D. in Computer Science at Arizona State University with Prof. Heni Ben Amor!Imperial College London: Invited Talk
[] I will be giving brief summary and outlook of my work presented in our NeurIPS 2020 paper at the Imperial College London!Resident @ X, The Moonshot Factory
[] Over the summer, I will be a resident at X, The Moonshot Factory, where I will be working on industrial manipulation tasks for Intrinsic, a robotics software and AI project at X .Video: Language Conditioned Imitation Learning
[] We contributed a video to the robot expo at IJCAI 2021 that is a direct extension to our NeurIPS 2020 paper. You can check out the video here!Paper: Language-Conditioned Imitation Learning for Robot Manipulation Tasks
[] We published a new paper at NeurIPS 2020! Our paper got accepted as a spotlight presentation (top ~4% of accepted papers). View full paperIntel AI Labs: Invited Talk
[] I am excited to give a talk, Language for Robotics at Intel AI Labs summarizing our efforts on learning robot policies from natural language instructions.Teaching Introduction to Theoretical Computer Science at ASU
[] I will be teaching CSE 355: Introduction to Theoretical Computer Science at Arizona State University as the main instructor during the upcoming Summer 2020 semester!Intel AI: Talk at the Deep Learning Community
[] I will be giving a talk at the Deep Learning Community of Practice titled Imitation Learning for Adaptive Robot Control Policies from Language, Vision, and Motion.Workshop Paper: Imitation Learning of Robot Policies by Combining Language, Vision and Demonstration
[] We contributed a workshop paper to the Workshop on Robot Learning at NeurIPS 2019!Best Poster Award
[] I received the Best Poster Award by Nvidia at the Southwest Robotics Symposium for my work on Neural Policy Translation for Robot Control!Robotics Intern @BOSCH
[] I will be joining Robot BOSCH in Sunnyvale for an internship to work on semantic data analysis with a focus on time series segmentation.Paper: Extrinsic Dexterity through Active Slip Control using Deep Predictive Models
[] We got our paper accepted to ICRA 2018, and I will be presenting our work in Brisbane, Australia! Paper on IEEE XploreBest Video Award
[] Awarded at the International Conference on Humanoid Robots (Humanoids) 2016 for our work on Learning human-robot interactions from human-human demonstrations Video on YouTube