Max Robotics — Global Service Robot Directory + US Certification

About

Marion Lepert is a leading researcher at the intersection of robotics, artificial intelligence, and human-robot interaction, whose work is defining how robots can assist humans in everyday environments. Her most impactful contribution is the development of **TidyBot**, a system that leverages large language models (LLMs) to enable robots to learn and generalize user preferences for household cleanup. This work, which has garnered over 270 citations across its variants, demonstrates how robots can personalize physical assistance by understanding natural language commands and applying learned preferences to novel situations—a critical step toward truly helpful home robots. Lepert is also a key contributor to the **DROID dataset**, a large-scale, in-the-wild robot manipulation dataset with over 100 citations, which provides the diverse, high-quality data essential for training robust manipulation policies. Her research further extends to tactile-informed manipulation, where she has developed action primitives like burrowing and excavating to help robots retrieve objects from dense clutter without jamming. Through these contributions, Lepert is not only advancing the technical capabilities of robotic manipulation but also shaping a future where robots can seamlessly integrate into human homes, learning and adapting to individual needs.