Researchers have developed GaitSpan, a novel framework designed to expand a basic humanoid walking policy into more dynamic locomotion such as jogging and running. This approach treats walking as a foundational skill that can be adapted by regenerating motor structures at new rhythms, extending stride length, and correcting motion through residual adaptation. GaitSpan is the first to enable a single command-conditioned policy for humanoids to cover a continuous speed range from walking to running, demonstrating transferability across different morphologies and terrains without additional training. AI
IMPACT Enables more versatile and adaptable humanoid robots capable of a wider range of locomotion speeds and terrains.
RANK_REASON Academic paper detailing a new AI framework for robotics. [lever_c_demoted from research: ic=1 ai=1.0]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
