Researchers Develop Solar-Powered Synaptic AI Device for Energy-Efficient Motion Detection

The innovative device mimics human synaptic behavior, enabling low-power, high-accuracy edge AI applications across diverse fields.

A team from Tokyo University of Science has created a dye-sensitized solar cell-based synaptic device for physical reservoir computing (PRC).
The device mimics human synaptic behavior, processing time-series optical data with adjustable time constants controlled by light intensity.
It achieves over 90% accuracy in classifying human motions like walking, running, and jumping, while consuming only 1% of the power required by conventional systems.
The technology integrates optical input, AI computation, analog output, and power supply functions, significantly reducing energy use and associated carbon emissions.
Potential applications include surveillance and car cameras, wearable medical devices, and smartwatches, with the promise of lower costs and enhanced energy efficiency.