Max Robotics — Global Service Robot Directory + US Certification

About

Avi Singh is a prominent robotics and machine learning researcher whose work sits at the intersection of deep reinforcement learning, robotic manipulation, and data-efficient learning systems. His research tackles one of the field's most pressing challenges: enabling robots to learn complex behaviors in the real world without exhaustive human supervision or reward engineering. Singh's most influential contribution, "End-to-End Robotic Reinforcement Learning without Reward Engineering" (2019, 208 citations), demonstrated how robots could learn directly from raw sensory inputs like camera images, eliminating the need for hand-crafted reward functions. This work helped democratize real-world robotic learning by making it significantly more practical. He further advanced the field through COG (2020) and "The Ingredients of Real-World Robotic Reinforcement Learning" (2020), exploring how offline reinforcement learning and accumulated past experience can accelerate the acquisition of new robotic skills. His work on data-driven behavioral priors through PARROT (2020) and sim-to-real transfer via i-Sim2Real (2022) demonstrates a consistent focus on sample efficiency and scalability. A compelling applied achievement is his involvement in high-speed robotic table tennis (2023), showcasing these principles in demanding real-world conditions. Across his career, Singh has consistently pushed reinforcement learning from controlled laboratory settings toward robust, deployable robotic systems.