Building General-Purpose Home Robots
Explore the latest advancements in building general-purpose home robots in this comprehensive talk by Lerrel Pinto, Assistant Professor of Computer Science at NYU. Delve into the concept of a "generalist machine" for domestic assistance, examining recent efforts to create adaptable and cost-effective robotic helpers. Learn how large, pretrained vision-language models can enhance mobile manipulation tasks, and discover the challenges in scaling beyond basic picking operations. Investigate the development of systems and algorithms for rapid skill acquisition, including new data collection tools, improved visual representations, and trial-and-error learning during deployment. Gain insights into learning approaches for multi-fingered hands that support dexterous behaviors and combine rich touch sensing with vision. Conclude by examining unsolved problems in the field and their potential impact on the future of general-purpose home robots.