Design for autonomous driving
Head D100
Overview
Autonomous Driving AI Semiconductor
The Head D100 is an AI semiconductor designed for Level 4 autonomous driving processing. It supports high-resolution multi-image sensors and is equipped with an ISP that has a sensitivity of over 140dB for accurate image recognition, as well as a high-performance NPU rated at 160 TOPS.
Key Features
High-Performance AI Accelerator
Built-in high-performance, low-power NPU with over 160 TOPS supporting various state-of-the-art artificial neural networks for autonomous driving.
Multi-Image Sensor Processing
Supports 8 MIPI camera inputs per chip (8 x 2MP, 2 x 8MP).
Image Signal Processor
Features 140dB class WDR (Wide Dynamic Range) performance for improved autonomous driving recognition and a multi-channel ISP to minimize frame delay.
3D Point Cloud Preprocessor
A preprocessing accelerator that receives LiDAR data for 3D Point Cloud deep learning.
Multi-Core CPU
Utilizes a 24-core CPU for preprocessing, post-processing, and various tasks.
Support for Various Interfaces
Supports Ethernet, CAN, PCIe, USB, HDMI, xSPI, SD/eMMC, and more.
Functional Safety
Supports ISO 26262 ASIL-B at the chip level and ASIL-D compliance at the system level.
Specifications
AI Performance
160 TOPS
Camera
8x MIPI CSI-2 RX
ISP
8x 2MP or 2x 8MP, 140dB WDR
CPU
Arm Cortex-A53, 24 cores
Memory
128-bit LPDDR5
Functional Safety
ISO 26262 ASIL-B
Process
12nm