On garment intelligence influenced by artificial neural networks and neuromorphic computing is emerging as a research direction in the e-textile sector. In particular, bio inspired Spiking Neural Networks mimicking the workings of the brain show promise in recent ICT research applications. Taking such technological advancements and new research directions driving forward the next generation of e-textiles and smart materials, we present a wearable micro Brain capable of event driven artificial spiking neural network computation in a fabric based environment. We demonstrate a wearable Brain SNN prototype with multi-layer computation, enabling scalability and flexibility in terms of modifications for hidden layers to be augmented to the network. The wearable micro Brain provides a low size, weight and power artificial on-garment intelligent wearable solution with embedded functionality enabling offline adaptive learning through the provision of interchangeable resistor synaptic weightings. The prototype has been evaluated for fault tolerance, where we have determine the robustness of the circuit when certain parts are damaged. Validations were also conducted for movements to determine if the circuit can still perform accurate computation.
|Publication status||Published - 25 Feb 2022|