World-Cognition Model Enhances Human-Robot Interaction
A new model called the World-Cognition Model (WCM) has been developed by researchers. This human-centered embodied agent utilizes the SLAK architecture, which stands for Sensing, Logic, Action, and Knowledge, and operates with an asynchronous runtime. WCM aims to fill the void in physical task interactions, where robots often lack user control and transparency. By distinguishing between perception, reasoning, control, and memory, the model facilitates simultaneous reasoning, dialogue, and execution. Notably, it features a human-in-the-loop teaching mode, allowing users to instruct robots on complex or long-term tasks interactively. Detailed findings are available in arXiv:2607.22999.
Key facts
- WCM is built on the SLAK architecture (Sensing, Logic, Action, and Knowledge).
- The model uses an asynchronous runtime for concurrent reasoning, dialogue, and execution.
- WCM introduces a human-in-the-loop teaching mode for interactive task instruction.
- The research is published on arXiv with ID 2607.22999.
- Current robot-control paradigms lack transparency and user interaction mechanisms.
- WCM separates perception, reasoning, control, and memory.
- The model aims to improve generalizable human-robot interaction in physical tasks.
- Teaching mode allows users to correct and redirect robot actions.
Entities
Institutions
- arXiv