Edge Computing and Digital Systems
The EDGE (Edge Computing and Digital Systems) team focuses on embedded systems, design and optimization of wireless sensor networks, system-on-chips and smart devices.
Research topics include optimization of energy consumption in smart devices, self-adaptive reconfigurable systems, reactive and cooperative systems, behavioral modelling, and RF (radio frequency) systems modelling. Furthermore, bio-inspired neuromorphic architectures design makes for a strong and original research subject to bring intelligence to these devices.
Goals are numerous and innovative. Modelling and optimized design of smart devices are performed through a systemic approach (including all the components of the device along with its environment), in order to strive for energy self-sufficient smart connected devices. Researchers follow a interdisciplinary approach, exploring new models from neurosciences to compare them with traditional machine learning methods. Promising models are implemented onto hardware architectures while optimizing energy efficiency in order to enable their deployment in constrained environments such as robotics, self-driving vehicles and IoT (Internet of Things).
Research axes
eWise
Energy-aware wireless sensor networks.
eSoC
Energy efficiency of SoC.
Transverse
Verification techniques based on synchronous approach.
Demonstrations
Simulation of a dynamic self-organizing map network (DSOM) for unsupervised learning (e.g. MNIST).
MicroAI project: Human Activity Recognition at the edge using smart connected glasses (with Ellcie-Healthy company).
Gesture recognition with spiking neural network and event-based camera on embedded neuromorphic hardware (SPLEAT).
Object detection with spiking neural network on embedded neuromorphic hardware (SPLEAT).