EyeQ
|
EyeQ™, Mobileye's System-on-Chip (SoC) delivers a solution for computationally intensive applications for real-time visual recognition and scene interpretation for use in intelligent vehicle systems. The key strengths of the EyeQ™ lie in its use of general CPUs (32 bit RISC ARM946E) and 4 patented Mobileye programmable ASIC cores - Vision Computing Engines (VCEs) - which are optimized for computing major time-consuming image processing tasks. All cores work in parallel to give the EyeQ unrivalled computing power in a small-package cost-effective processor specifically designed for the automotive vision based Driver Assistance Systems (DAS) market. Mobileye EyeQ™ has been launched on multiple major US and European platforms to date, covering applications of lane departure warning (LDW), LDW with vehicle detection with fusion (active braking), traffic sing recognition and more. Lane Departure Waning: LDW & Vehicle Detection & Radar-Vision Fusion with Collision Mitigation by Braking: LDW & Intelligent High Beam control & Traffic Sign Recognition: Moving forward Mobileye will launch EyeQ based systems for additional Vehicle Makers in the Far East in 2009 with Lane Departure Warning and 2010 for a major European vehicle maker with Lane Departure Warning and Intelligent High Beam control. Since 2007 the Mobileye EyeQ™ chip has also been used in the Mobileye's aftermarket road safety systems, which features Lane Departure Warning & Vehicle Detection providing full speed range Forward Collision Warning & Headway Monitoring and Warning. Mobileye EyeQ™ is manufactured by STMicroelectronics. For a detailed EyeQ product brief, contact Sales@mobileye.com The chip architecture is designed to run a full-fledged application on a single chip, and is completely programmable to accommodate a wide range of visual processing applications beyond automotive specific applications. 
Mobileye EyeQ™ is manufactured using the leading CMOS 0.18-micron technology, operating at 120Mhz. To optimize cost performance, all peripheral interfaces are integrated in to the SoC, including dual CAN Controllers; a UART, SDRAM controller, parallel I/O, and Video image data capture units. EyeQ™ is an ultra low cost device but with the theoretical equivalent computation power of an Intel Pentium IV processor, running at 4Ghz clock rate. The Mobileye EyeQ™ architecture consists of two 32 bit RISC ARM946E CPUs, four Vision Computing Engines (VCE), a multi channel DMA and several peripherals. The ARM946E Control CPU manages the four VCEs, the ARM946E Logic and the multi-channel DMA as well as the other Peripherals. The four VCEs and the ARM946E Logic perform all the intensive vision computations required by the applications such as tracking and pattern classification. The Multi-Layer 4x4 Matrix offers a high connectivity scheme that is needed for providing the required data bandwidth of the vision processing. The Multi-Layer matrix routes the four master ports to the four slave ports and enables concurrent operation of up to four AHB buses. If there is a bus contention on a slave port, the Multi-Layer AHB matrix decides on the winning AHB master according to the priority scheme. All the VCEs work in parallel, retrieving their tasks from tasks queues (one per each VCE). The VCEs communicate over the high bandwidth Multi-Layer Matrix block, via a common master port. The task queues are managed by the ARM9E Control CPU. A high-speed, 64bit width, 288Kbyte on-chip SRAM is located on this Multi-Layer Matrix for fast image memory storage and retrieval. A separate 32-bit low bandwidth Peripheral Bus (VPB) is provided to connect all of the various peripherals such as the CAN Controllers.
Vision Computing Engines (VCEs) Classifier Engine: Tracker Engine: Lane Detection Engine: Window, Preprocessing and Filter (WPF) Engine: DMA Controller: Main features are: Multi Layer 4x4 Matrix: Video Interface: Memory support: Other peripherals: Serial Interfaces:
|
