Neuromodulation based control of autonomous robots

The paper presents a control approach based on neuromodulation in vertebrate brains and its implementation on an autonomous robotic platform. The neuromodulatory function is modeled through a neural network for generating context based behavioral responses to sensory input signals from the environment. Three types of neurons are incorporated in the neural network model. The neurons are- cholinergic and noradrenergic (ACh/NE) for attention focusing and action selection, dopaminergic (DA) for curiosity-seeking, and serotonergic (5-HT) for risk aversion behaviors. The neuronal model was implemented on a relatively simple autonomous robot that demonstrated its interesting behavior adapting to changes in the environment.