0 views collected around this technical thread.
This article describes how Didi’s vehicle‑vision team built an ultra‑low‑power ADAS solution on the JueShi dash‑cam platform, using lightweight detection models, temporal fusion, camera‑calibration techniques and data‑driven optimization to cut rear‑end collision rates by over 11% and improve overall traffic safety.
DiDi’s ultra‑low‑power JueShi ADAS combines lightweight vision models, temporal‑fusion Kalman filtering, and camera‑calibration techniques to deliver real‑time forward‑collision warnings and brake‑light alerts, cutting rear‑end crashes by over 11% and overall accidents by 9% through continuous edge‑AI learning.
The article explains how the memory‑wall limitation of traditional von Neumann architectures hampers AI chip performance, describes two in‑memory computing approaches—circuit‑level modifications and new memory devices—highlights recent conference trends, and showcases a Chinese startup’s 8‑bit low‑power in‑memory AI chip that could enable ubiquitous AI on edge devices.
Prof. Yin Shouyi of Tsinghua University presented a reconfigurable, low‑bit quantized neural‑network architecture and the Thinker‑I, Thinker‑II, and Thinker‑S chips, demonstrating ultra‑low power consumption and high energy‑efficiency for AI deployment on edge devices.