MIT EEMS Group - PI: Vivienne Sze

Emerging fields such as artificial intelligence (AI) and robotics rely on computationally complex algorithms to achieve high accuracy that enable new applications and discoveries. While much of the compute for these applications currently happens in the cloud, processing locally on the device is often desirable due to latency, security/privacy, and communication bandwidth concerns. However, local processing on an embedded device is challenging as the energy is limited by its battery capacity.

We aim to drastically reduce the energy consumption of these computationally complex algorithms for embedded applications while delivering real-time performance. We develop efficient computing methods that cross traditional research boundaries between algorithms and hardware. Our approaches have been demonstrated in a broad range of applications including deep neural networks, autonomous navigation for robotics, and video compression. 

Info about accessibility  at https://accessibility.mit.edu/