Organic Neuromorphic Robot Capabilities in Learning Through Experience

Organic Neuromorphic Robot Capabilities in Learning Through Experience

Different scientists working with Paschalis Gkoupidenis—Paul Blom’s department leader of Max Planck Institute for Polymer Research in collaboration with Universities of Eindhoven, Stanford, Brescia, Oxford, and KAUST, steered a robot through the maze using an organic neuromorphic circuit as the result of the application of the basic principle of learning through experience. Krauhausen, a doctoral student at TU Eindhoven, wanted to use the simple setup to show how powerful the organic neuromorphic device is in the real world.

The researcher fed the smart adaptive circuit with sensory signals coming from the environment to achieve the robot’s navigation inside the maze. The maze’s path toward the exit is indicated visually at each maze intersect where the robot often misinterprets the visual sign that results in the robot making a wrong “turning” at the maze intersects and losing its way out. The robot taking wrong turns and making wrong decisions that end in wrong dead-end paths can be prevented by receiving corrective stimuli. For example, when the robot hits a wall, the corrective stimuli are directly applied to the organic circuit via electrical signals induced by a touch sensor attached to the robot. With each subsequent execution of the experiment, the robot learns to make the right “turning” decisions at the intersects to avoid receiving corrective stimuli. After a few trials, the robot finds its way out of the maze by gradually learning. This learning through experience process happens exclusively on the organic adaptive circuit.

Gkoupidenis said that they are glad to see that the robot can pass through the maze after learning a simple organic circuit from a very simple setup. Gkoupidenis also indicates their hope in the organic neuromorphic devices for their future usage for local and distributed learning, where it will open different opportunities for real-world robotics, human-machine interface, and point-of-care diagnostic applications. Emerging novel platforms for rapid prototyping and education with the intersection of materials science and robotics are also expected.

Follow this Link!

- Fun facts and Trivia brought to you by the Research and Development Committee -