Add to EJ Playlist Summary of the bachelor thesis project by Manuel Watter. Using Neural Fitted Q-Iteration, a robot head first learns to keep a stationary object centered in the camera, and then to adjust its actions so as to better keep it in view once it is moving.
Add to EJ Playlist Learning a controller for the Ms. Pac-Man arcade game by trial and failure. The learning algorithm is TD lambda using a neural network for value function representation and advanced features as input.
Add to EJ Playlist Showcase of the prototype system developed during the first stage of the NeuroBots project (https://www.br ainlinks-braint ools.uni-freibu rg.de/research/ projects/projec t-details/neuro bots) as a collaboration between the iEEG Lab (http://www.iee g.uni-freiburg. de) and the Machine Learning Lab (http://ml.info rmatik.uni-frei burg.de) of the University of Freiburg. Imagined motor commands are used for high-level remote control of an autonomous, reinforcement-l earning-based robotic system for reaching and grasping several kilometers away.
Add to EJ Playlist Teaser video for the BrainLinks-Brai nTools Cluster of Excellence (https://www.br ainlinks-braint ools.uni-freibu rg.de/). Improving on our previous system (http://www.you tube.com/watch? v=ZIQ6uyptAB8), we use electroencephal ography to communicate higher-level intentions to an autonomous controller learned via Neural Fitted Q-Iteration.
Add to EJ Playlist First attempt at combining electrooculogra phy and machine learning to have a Katana robotic arm search for an object and pick it up.
Collaboration between the Machine Learning Lab and the Biomedical Microsystems Group.
Add to EJ Playlist A Kinova Jaco robotic arm is expected to learn how to reach for an object and pick it up autonomously. Control is realized in a visual feedback control loop, making it both reactive and robust to noise. The controller is learned from scratch, without prior knowledge of proper behaviour, by success or failure using Neural Fitted Q Iteration.