Coursera Flash Sale
40% Off Coursera Plus for 3 Months!
Grab it
Explore the cutting-edge field of humanoid robot learning in this 56-minute conference talk that examines the unique challenges and promising solutions for developing general-purpose humanoid intelligence. Discover how humanoid robots represent the ideal physical embodiment for assisting with diverse daily tasks in human-centric environments, driven by hardware advancements, AI progress, and growing automation demands. Learn about the specific challenges humanoids pose in robot learning settings and examine innovative approaches to scale learning through novel tools including HumanoidBench and MuJoCo Playground. Understand flexible algorithms such as OmniRetarget and FastTD3, along with expressive architectures like the Body Transformer that advance humanoid capabilities. Gain insights into research focused on enhancing humanoid robots' intelligence and loco-manipulation abilities by incorporating priors, inductive biases, and multi-sensory feedback. The presentation covers vision-based, data-driven tactile sensors and their applications in robot control and dexterous manipulation, drawing from award-winning research including the Best Demo Paper Award at RSS 2025, Best Paper Award at RoboSoft 2022, and recognition as a Rising Star at RoboSoft 2025.
