A full-hardware edge detection system based on a highly robust dual random number generator.

A 32-channel parallel full-hardware edge detection system using dual random number computation with 64 Ag/AlO x /n-Si memristors achieves 31.25 frames per second with strong anti-interference.

• Why it matters: Edge computing for autonomous driving needs low-power, interference-resistant, real-time image processing, but full-hardware edge detection remains difficult under chip area and power constraints.
• What they did: The work fabricated a volatile Ag/AlO x /n-Si memristor and used electron microscopy to visualize Ag conductive filament formation, then integrated two memristors with logic for dual random number hardware computation.
• The result: The edge unit consumed 92.3% less energy than traditional logic circuits and, in a 32-channel parallel system with 64 devices, delivered 31.25 frames per second with excellent edge detection.

AI summary of the abstract · may contain errors · not medical advice · read the paper.