Neuromorphic Computing

Our research activities center on improving the learning capability and intelligence of technical systems, whether in the manufacturing, automobile and robotics. To carry out these activities, we rely on findings from the field of neurobiology and apply software methods from the area of artificial intelligence (AI) and the subarea of deep learning. We also develop algorithms and software for highly energy-efficient neuromorphic hardware and its use in the area of machine learning.            

1. Spiking Adaptive Locomotion Control on Neuromorphic Hardware
   a) In the frame of the Human Brain Project and in collaboration with the EPFL biorobotics lab, we have implemented a novel spiking algorithm to control bio-inspired robots using neuromorphic hardware. To reach our goal, we use the Neurorobotics Platform, a simulation tool that we co-develop and that enables us to embody our control algorithms in many different experimental scenarios.
   b) Together with our partner Intel, we research a low-latency spiking robotic arm controller to enable for fast object insertion. A hardware demonstrator is being developed.
    
2. Spiking Optical Flow with Event-Based Cameras
   Optical flow is an important area of research in image processing, which is used to estimate movements within a visual scene. Event-based cameras are the ideal neuromorphic hardware for capturing visual scenes and processing them with spiking neural networks, eliminating the inherent latency of frame grabbers. Our field of competence conducts research into methods for efficient spiking optical flow computations and their use in industrial applications such as mobility and automation. Here as well, the Neurorobotics Platform is an essential tool for training in simulation.

Neuromorphic hardware