Coursera Flash Sale
40% Off Coursera Plus for 3 Months!
Grab it
Explore a scalable framework for developing humanoid foundation models in this one-hour conference talk. Learn about two key developments that address the challenge of generating large-scale training data for humanoid robots without relying heavily on teleoperation demonstrations. Discover how an optimization-based task and motion planning (TAMP) framework generates diverse strategies for accomplishing different tasks while leveraging pre-trained vision-language models to propose manipulation sequences as subgoals. Understand the process of retargeting vast amounts of human motion data to humanoid robots and see how these data sources combine to build effective foundation models. Gain insights into the practical applications of these approaches and conclude with an overview of safety considerations in humanoid robotics research.
