I am a PhD student at the Computer Science and Artificial Intelligence Laboratory (CSAIL) at MIT, advised by Pulkit Agrawal. I study learned control as a component of complete robotic systems. Previously, I received my BS ('20) and MEng ('21) degrees at MIT.
Research
SoftMimic: Learning Compliant Whole-body Control from Examples
Gabriel B. Margolis*, Michelle Wang*, Nolan Fey, Pulkit Agrawal
Preprint, 2025
Variable-stiffness humanoid motion imitation enabling gentle contact with the environment for safety and generalization.
Bridging the Sim-to-Real Gap for Athletic Loco-Manipulation
Nolan Fey, Gabriel B. Margolis, Martin Peticco, Pulkit Agrawal
Robotics: Science and Systems (RSS), 2025
Improve the sim-to-real transfer of a legged manipulator's athletic motor skills by aligning the physics of simulation to reality through a data-driven approach.
Position: Automatic Environment Shaping is the Next Frontier in RL
Younghyo Park*, Gabriel B. Margolis*, Pulkit Agrawal
International Conference on Machine Learning (ICML), 2024 (Oral Presentation)
To generalize the successes of reinforcement learning, we need research on the interaction between environment shaping and learning dynamics.
Action Space Design in Reinforcement Learning for Robot Motor Skills
Julian Eßer*, Gabriel B. Margolis*, Oliver Urbann, Sören Kerner, Pulkit Agrawal
Conference on Robot Learning (CoRL), 2024
Examine the mechanisms by which action space design impacts learning dynamics across robot embodiments, including evoBOT, a unique wheeled-legged biped.
Learning Force Control for Legged Manipulation
Tifanny Portela, Gabriel B. Margolis, Yandong Ji, Pulkit Agrawal
International Conference on Robotics and Automation (ICRA), 2024
Learn to control the force at the end effector of a quadruped robot with an arm for compliant and forceful loco-manipulation.
Maximizing Quadruped Velocity by Minimizing Energy
Srinath Mahankali*, Chi-Chang Lee*, Gabriel B. Margolis, Zhang-Wei Hong, Pulkit Agrawal
International Conference on Robotics and Automation (ICRA), 2024
Balance energy and velocity through constrained optimization for agile running with minimal reward shaping.
Learning to See Physical Properties with Active Sensing Motor Policies
Gabriel B. Margolis, Xiang Fu, Yandong Ji, Pulkit Agrawal
Conference on Robot Learning (CoRL), 2023
Learn to see how terrains feel by collecting self-supervised data with information-maximizing motor skills.
DribbleBot: Dynamic Legged Manipulation in the Wild
Yandong Ji*, Gabriel B. Margolis*, Pulkit Agrawal
International Conference on Robotics and Automation (ICRA), 2023
Dynamic object manipulation with the legs, using onboard computation and sensing.
Walk These Ways: Tuning Robot Control for Generalization with Multiplicity of Behavior
Gabriel B. Margolis, Pulkit Agrawal
Conference on Robot Learning (CoRL), 2022 (Oral Presentation)
One learned policy embodies parameterized dynamic behaviors useful for different tasks.
Rapid Locomotion via Reinforcement Learning
Gabriel B. Margolis*, Ge Yang*, Kartik Paigwar, Tao Chen, Pulkit Agrawal
Robotics: Science and Systems (RSS), 2022
High-speed running and spinning on diverse terrains with a single neural network.
Learning to Jump from Pixels
Gabriel B. Margolis, Tao Chen, Kartik Paigwar, Xiang Fu, Donghyun Kim, Sangbae Kim, Pulkit Agrawal
Conference on Robot Learning (CoRL), 2021
A hierarchical control framework for dynamic vision-aware locomotion.