Multimodal Embodied Agent
A goal-directed intelligent system that combines multimodal perception, task reasoning and physical action execution and can revise decisions based on environmental feedback, subject in the real physical world to the irreversible consequences of actions.
Timeline (last 90 days)
The paper proposes the unified PAPAV capability framework, comprising five core capabilities: Perceive, Anticipate, Plan, Act, and Verify.
Multimodal embodied agents (MMEAs) are defined as goal-directed systems that combine multimodal task reasoning with physical actions and revise decisions based on feedback.
All Facts (2)
The paper proposes the unified PAPAV capability framework, comprising five core capabilities: Perceive, Anticipate, Plan, Act, and Verify.
50%UnverifiedMultimodal embodied agents (MMEAs) are defined as goal-directed systems that combine multimodal task reasoning with physical actions and revise decisions based on feedback.
50%