An embodied model for sensorimotor grounding and grounding transfer: experiments with epigenetic robots

Area of Science:

• Cognitive Science
• Robotics
• Artificial Intelligence

Background:

• Symbol grounding is crucial for psychologically plausible computational models of language.
• Epigenetic robotics offers a novel approach to modeling autonomous mental development.

Purpose of the Study:

• To present an embodied computational model for grounding language in action using epigenetic robots.
• To demonstrate how linguistic abilities can be integrated with and dependent on other behaviors and skills.

Main Methods:

• Utilized simulated robots learning action names through imitation.
• Employed grounding transfer to enable learning of higher-order action concepts.
• Developed an integrative model where language is dependent on grounded behaviors.

Main Results:

• Simulated robots successfully learned basic action names via imitation.
• Grounding transfer facilitated the acquisition of higher-order action concepts.
• The model showed autonomous development of new behavioral abilities from grounded categories.

Conclusions:

• Embodied, action-based grounding is fundamental for developing psychologically plausible language models.
• Epigenetic robotics provides a viable framework for autonomous learning and development of complex behaviors.
• Linguistic interaction can drive the emergence of higher-order cognitive and behavioral skills.