Neuromorphic Retina Chip Accelerates Motion-Based Vision

Shuo Gao and colleagues describe a neuromorphic, retina-inspired chip in a recent paper that rapidly detects motion and focuses vision algorithms. The hardware-accelerated system reduced processing latency to roughly 100 microseconds and sped up vision pipelines by about 400 percent, improving robotic-arm tracking by over 740 percent and integrating with optical flow and YOLO for vehicles and robots.