Max Robotics — Global Service Robot Directory + US Certification

About

Todor Davchev is a robotics researcher specializing in robot learning, manipulation, and imitation learning, with a particular focus on enabling robots to acquire complex skills efficiently and generalize across diverse tasks. His most influential contribution, "Residual Learning From Demonstration" (2022, 48 citations), introduced a framework combining Dynamic Movement Primitives with residual learning to tackle contact-rich manipulation tasks such as peg-in-hole insertions — a long-standing challenge due to the complexities of friction and contact dynamics. This work, developed across multiple iterations, demonstrated a principled approach to adapting classical motor primitives through learned corrections. His research on "RoboTAP" (2024, 31 citations) pushed the frontier of few-shot visual imitation, using point tracking to enable rapid onboarding of new robot tasks without task-specific engineering. His earlier "Vid2Param" series explored extracting dynamical parameters directly from video, bridging perception and physical reasoning for robotic systems. Contributing to the ambitious "RoboCat" project (2023), Davchev also engaged with generalist, self-improving multi-task robotic agents inspired by foundation models. Collectively, his work advances the goal of making robots faster, more adaptable, and practically deployable beyond controlled laboratory settings.