Max Robotics — Global Service Robot Directory + US Certification

About

Devin Schwab is a robotics researcher whose work sits at the intersection of reinforcement learning, computer vision, and multi-agent systems, with a particular focus on real-world robot control and transfer learning. His most impactful contributions center on enabling robots to learn complex manipulation and locomotion skills directly from vision, without the need for extensive hand-coding or simulation-to-reality fine-tuning. In his highly cited 2019 paper, "Simultaneously Learning Vision and Feature-Based Control Policies for Real-World Ball-In-A-Cup" (15 citations), Schwab introduced a novel multi-task reinforcement learning framework that trains policies across different state-spaces—vision and feature-based—simultaneously, dramatically accelerating real-world policy learning. He has also pioneered zero-shot transfer learning for multi-agent systems, as demonstrated in his 2018 work (12 citations), where policies trained with a fixed number of agents can be deployed without retraining in environments with varying team sizes and opponent counts—a critical capability for domains like RoboCup, where robot breakages are common. Schwab's work on learning primitive skills for mobile robots (9 citations) further advances the goal of creating general-purpose, adaptable robot controllers. With a total of over 50 citations across his key publications, Schwab is establishing himself as a leading voice in practical, scalable reinforcement learning for robotics.