Max Robotics — Global Service Robot Directory + US Certification

About

Bowen Baker is a leading researcher in reinforcement learning and robotics, whose work has significantly advanced the capabilities of autonomous systems. His primary research areas include dexterous manipulation, multi-goal reinforcement learning, and learning from video data. Baker's most impactful contribution is his pioneering work on "Learning dexterous in-hand manipulation" (2019, 1,588 citations), where he demonstrated that complex, vision-based object reorientation on a physical Shadow Dexterous Hand could be learned entirely through reinforcement learning in simulation, using domain randomization to bridge the sim-to-real gap. This breakthrough opened new possibilities for robotic manipulation. He also co-developed the "Multi-Goal Reinforcement Learning" suite (2018, 196 citations), introducing a widely-adopted set of challenging robotics benchmarks for pushing, sliding, and pick-and-place tasks. More recently, Baker has explored learning from internet-scale data with "Video PreTraining (VPT)" (2022, 50 citations), a method that enables agents to learn to act by watching unlabeled online videos, promising to unlock general-purpose decision-making for robotics and beyond.