An Environment Cognition and Motor Adaptation Model Eliciting Sensorimotor Constraints Based on Time-Series Observations

摘要

Regardless of complex, diverse, dynamically changing environments, animals recognize situated worlds and respond appropriately through interaction with the environment. We investigated sentient environment cognition and motor adaptation, focusing on "automatic movement". Thanks to periodicity, environment cognition and motor adaptation of body movement to environmental perturbations is modeled relatively easily. We propose an adaptive neural oscillator model inspired by biological brainervous system for periodic motion control of physical controlled objects such as pendulums. A key concept for realizing environment cognition and motor adaptation is the context-based elicitation of sensorimotor constraints canalizing suitable periodic movement. Results of control tasks demonstrated the feasibility of our proposal.