PhD ceremony Thorben Schoepe: Uncovering the potential of neuromorphic sensorimotor systems

On the quest of understanding the brain

Animals stand in constant interaction with their environment, searching for food, fleeing from a predator or looking for a mating partner. Evolution has optimized these organisms over the last 500 million years. Therefore, animals may serve as a role model for new inventions such as autonomous cars, drones and vacuum cleaner robots.

In order to develop systems inspired by nature we need to understand how animals function. To this end, Thorben Schoepe built neuromorphic robots: robots that imitate a natural brain. He let them perform various tasks including sound source localization, obstacle avoidance and straight line navigation. His results provide many new hypothesis about animal cognition. For example, how time scales of decision making could be implemented, how a mixture of goal directed movement and random exploration help finding a pathway and how global optic flow controls the movement velocity.

On the neural level, Schoepe shows that spike coding can be useful to encode information in various different ways including population coding, rate coding, spatial coding and temporal coding. He further shows that insect-like behaviour can emerge from the complex recurrent and massively parallel interplay of a few thousand neurons and hundred thousand synapses. These results successfully demonstrate on a low level of abstraction how insects and vertebrates interact with the environment. Only very few other fully spiking closed-loop systems have been developed so far. Thus, this thesis might serve as a starting point for exploring the functionality of embodied cognition in fully spiking neuromorphic sensorimotor systems.

The ceremony can be followed via this livestream. For more information see the PhD ceremony page.