Max Robotics — Global Service Robot Directory + US Certification

About

Misha Denil is a machine learning researcher specializing in data-driven robotics, reinforcement learning, and imitation learning, with a particular focus on enabling robots to learn complex manipulation tasks from experience and demonstrations. His most influential contribution is a scalable framework for data-driven robotics that combines large datasets of recorded robot experience with learned reward functions, allowing systems to generalize across multiple object manipulation tasks on real hardware — work that has accumulated over 100 citations across its various publications. Denil has made notable advances in reward learning, including positive-unlabeled reward learning to combat reward model exploitation, and adversarial imitation learning, where his research exposed critical vulnerabilities arising from task-irrelevant visual features. His work on the Intentional Unintentional agent extended deep deterministic policy gradients to handle simultaneous multi-task continuous control. More recently, he has contributed to generalist robotic agents through RoboCat, a self-improving system inspired by foundation models, and explored vision-language models as scalable success detectors for agent training. Across his career, Denil's research consistently pushes toward robots that learn efficiently, generalize broadly, and acquire new skills with minimal supervision.