Max Robotics — Global Service Robot Directory + US Certification

About

Arthur J. Petron is a leading researcher in robotic manipulation, reinforcement learning, and simulation-to-real transfer. His most influential work, "Learning dexterous in-hand manipulation" (2019, 1,588 citations), pioneered the use of reinforcement learning to train a Shadow Dexterous Hand to reorient objects using only vision, with policies learned entirely in simulation through randomized physical properties. This breakthrough was extended in "Solving Rubik's Cube with a Robot Hand" (2019, 632 citations), where Petron and his team demonstrated that a model trained solely in simulation could solve a Rubik's Cube on a real robot—a feat of unprecedented complexity. This was enabled by automatic domain randomization (ADR), a novel algorithm that systematically varies simulation parameters to bridge the sim-to-real gap. Petron's work has fundamentally advanced dexterous manipulation, making it possible to train complex, real-world robotic skills without physical hardware. His earlier research on multi-material 3-D viscoelastic modeling of transtibial residua (2016) reflects a broader interest in biomechanics and medical robotics. With over 2,300 total citations, Petron's contributions continue to shape the future of autonomous manipulation and embodied AI.