Neuromodulation, attention and localization using a novel Android™ robotic platform

摘要

We present a novel neural architecture based on neuromodulated attentional pathways. The neural network controlled a robot, which had to perform a reversal learning task based on GPS locations. We developed a robotic platform that leverages smartphones technology. The behavior of the robot was entirely driven by a neural network that ran on an Android phone, which handled sensor input from the phone and controlled the motor and servo of the robot. The robot managed to perform the task successfully by increasing attention to relevant locations and decreasing attention to irrelevant ones.