Max Robotics — Global Service Robot Directory + US Certification

About

Kyle Hatch is a leading researcher in robot learning, with a focus on scaling data-driven manipulation to real-world settings. His most impactful contribution is **DROID**, a large-scale, in-the-wild robot manipulation dataset that has rapidly garnered over 100 citations since its 2024 release. This work directly tackles the field’s core bottleneck: the lack of diverse, high-quality training data. By collecting data across numerous environments and robots, DROID provides a crucial foundation for training more robust and generalizable policies. Hatch also co-developed the **Train Offline, Test Online** benchmark, which systematically addresses the prohibitive costs and lack of standardization that have historically limited progress in robotics research. His work on **D5RL** further advances the field by curating diverse datasets for offline reinforcement learning, enabling methods that learn from pre-collected data without costly real-world exploration. More recently, Hatch has explored hierarchical control with generative models, using pretrained image models to plan intermediate subgoals for low-level policies. Through these efforts, Hatch is helping to democratize robot learning and pave the way for more capable, adaptable robotic systems.