Researchers have introduced the World-Cognition Model (WCM), an embodied agent designed to improve human-robot interaction for physical tasks. Built on the SLAK architecture, WCM separates perception, reasoning, control, and memory, allowing for concurrent operation of these components. A key feature is its human-in-the-loop teaching mode, which enables users to guide robots through complex or long-horizon tasks, refining the model through chain-of-thought supervision. In real-world tests, WCM achieved a 73.8% success rate across nine tasks, including those not used in its fine-tuning. AI
IMPACT This model could significantly improve the usability and teachability of robots in real-world physical tasks.
RANK_REASON The cluster contains an academic paper detailing a new model and architecture for human-robot interaction. [lever_c_demoted from research: ic=1 ai=1.0]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
