Max Robotics — Global Service Robot Directory + US Certification

About

Marcin Andrychowicz is a prominent machine learning researcher whose work sits at the intersection of deep reinforcement learning, robotics, and sim-to-real transfer. Best known for his landmark 2019 paper "Solving Rubik's Cube with a Robot Hand" (632 citations), he helped pioneer automatic domain randomization (ADR), a technique that enables policies trained entirely in simulation to transfer successfully to complex real-world robotic systems — a longstanding challenge in the field. Andrychowicz has made foundational contributions to sample-efficient reinforcement learning, most notably through "Hindsight Experience Replay" (HER, 352 citations), an elegant algorithm that dramatically improves learning from sparse, binary rewards by reframing failed experiences as useful training signal. This work has become a standard tool in goal-conditioned RL research. He further advanced the field with "One-Shot Imitation Learning" (229 citations), enabling robots to generalize tasks from a single demonstration, and helped establish benchmark robotics environments through the widely adopted OpenAI Gym-integrated multi-goal framework (196 citations). Across his body of work, Andrychowicz has consistently tackled some of reinforcement learning's hardest practical problems — sparse rewards, unsafe exploration, and sim-to-real generalization — leaving a substantial and lasting imprint on modern robotics research.