Sensorimotor Norm

Sensorimotor norms research investigates how organisms integrate sensory information with motor actions to achieve goals, focusing on understanding and replicating this ability in robots. Current research emphasizes developing models, such as transformers and recurrent neural networks, that learn from sensorimotor data through techniques like next-token prediction, predictive coding, and reinforcement learning, often incorporating elements of active inference and embodied cognition. This work aims to improve robot adaptability, generalization, and control, ultimately contributing to advancements in artificial general intelligence and human-robot interaction.