Neuromorphic Cognitive Robots

The NCR group develops neural architectures for neuromorphic hardware that can control autonomous robots, equipped with event-based neuromorphic, as well as conventional sensors.  The neural architectures are based on models of biological neural circuits and their robotic embodiment sheds light on how cognitive processes unfold in neural systems in a closed sensorimotor loop. The behaviours that our robots exhibit, while being controlled solely by mixed signal analog / digital neuromorphic devices, range  from simple reactive obstacle avoidance and target acquisition to cognitive behaviours like map formation or sequence learning.  

1. Sequence learning in neuromorphic systems (bird song learning, serial order, behavioral organisation).
2. State estimation and map formation with spiking neural networks on chip ("Neuromorphic SLAM")
3. Embodied long term memory: neuronal mechanisms and neuromorphic realisation (SNSF Ambizione Project "ELMA: Embodied Long-Term Memory for Cognitive Neuromorphic Agents")
4. Neuromorphic motor control (learning forward and inverse models in spiking neural networks on chip; Project with Intel using neuromorphic device Loihi)
5. Event-based vision architectures (active stereo vision with event-based sensors DVS; object tracking on SCAMP; event-based pattern / place recognition)
6. EU FETPROACT Coordination and Support Action NEUROTECH (http://neurotechai.ch)

• Alpha Renner
• Raphaela Kreiser
• Julien Martel

• Gwendolyn English
• Eloy Barrero
• Llewyn Salt
• Jianlin Lu (RWTH, Aachen)
• Ammar Bitar (University of Freiburg)
• Jonathan Müller
• Kay Müller
• David Niederberger
• Valery Metry
• Sebastian Glatz

Alexander Dietmüller and Herman Blum

Mario Blatter

Matteo Cartiglia

Sebastian Glatz

Nicolas Känzig

Panin Pienroj

Lukas Blässig

Lennard de Graf (TUM)

Michel Frising

Zahra Farsijani

Lin Jin

Viviane Yang, Paul Joseph, and Balduin Dettling

Davide Plozza and Damiano Steger

Héctor Vazquez (MIT)

Michael Purcell

Nuria Armengol

Jozef Bucko