Max Robotics — Global Service Robot Directory + US Certification

About

Norman Di Palo is at the forefront of a new wave in robotics, pioneering the use of foundation models—large language models (LLMs) and vision transformers—to create more intelligent, sample-efficient, and generalizable robot learning systems. His research centers on imitation learning, in-context learning, and safe human-robot interaction, with a core focus on enabling robots to learn complex manipulation tasks from minimal human demonstrations. Di Palo’s major contributions include demonstrating that LLMs can act as zero-shot trajectory generators (46 citations), challenging the assumption that they are only suitable for high-level planning. He also introduced DINOBot (26 citations), a novel imitation learning framework that leverages DINOv2 features for robust visual retrieval and alignment, and developed a method for in-context imitation learning using off-the-shelf text-based transformers (20 citations). His work on learning multi-stage tasks from a single demonstration and the SAFARI algorithm for safe active imitation learning further highlights his impact. With over 160 total citations and recent contributions to the Gemini Robotics project, Di Palo is a rising leader in building unified, foundation-model-driven robotic agents that bring AI into the physical world.