Neuromorphic controller for an autonomous quadcopter
A recently developed in our lab quadcopter is equipped with a neuromorphic camera (“artificial retina”) and a neuromorphic processor featuring spiking neurons with on-chip learning implemented in hardware. Using this hardware, neuronal controllers for avoiding obstacles, following targets, and mapping of the environment will be developed. The most proximal goal is to implement a simple obstacle-avoidance strategy based on a neurally-inspired “looming detector” and a simple reinforcement-learning architecture.
Starting date: on request
Background in computer science, electrical engineering, or mechanical engineering with interest in neural networks and brain-inspired robot control