GAPPoc : A Family of GAP8-centric Proof Of Concept boards for Edge AI

时间： 2024-08-14 来源：GREENWAVES Official Website

技术问答

GREENWAVES GAP8 application processor chip is great at analyzing and understanding data from IoT sensors, from the simplest to the most complex, in a very tight power envelope – from a few tens of milliwatts in active mode down to a few microwatts in sleep mode.

To evaluate the performance that can be expected from GREENWAVES GAP8 chip, GREENWAVES offers GAPuino, an evaluation board compatible with the Arduino ecosystem. This is a generic board that runs off some external power supply or USB and, depending on evaluation intent, can be augmented by additional « shields » (in Arduino terminology) bearing e.g sensors or communication modules.

To go beyond the generic evaluation enabled by GAPuino, GREENWAVES has also developed the « GAPPoc » concept. GAPPoc stands for GAP8 Proof of Concept. The idea here is to focus on a specific class of applications and embed on a single, compact board everything that is needed to demonstrate the target application: not only the GAP8 chip and closely coupled hardware such as crystal or external memory but also sensors and radio fit for that particular class of applications, as well as a battery.

GAPPoc boards are intended to be used as credible demonstrators representative of a final application in terms of features, performance, power profile and battery life as well as, up to a certain point, physical dimensions.

GAPPoc Architecture and Implementation Choices

All GAPPoc are architectured around a GAPMod core module, which itself essentially consists of a GAP8 chip and external memory (Flash+RAM) plus a low-power crystal oscillator and power management functions. The GAPMod core module (visible at the center of Fig.1) is assembled on GAPPoc like any other SMD device and is complemented with functionalities in line with the target application (especially sensor and radio) as well as generation of all power supplies from the on-board battery, plus various expansion/monitoring features.

With this approach, the « invariant » part of a GAP8-centric system, which also happens in many cases to be the most critical in terms of PCB design (with memory interfaces and sensitive clock generation hardware) is concentrated on the pre-built GAPMod core module, which can be reused across different variants of GAPPoc targetting different classes of applications.

In the same spirit, other implementation choices are made to keep the design of GAPPoc boards relatively simple and cost-efficient for prototype volumes (for instance, using pre-certified radio modules rather than custom designs on PCB). Also, despite a specific GAPPoc targetting a very specific class of applications, GREENWAVES nevertheless strives to keep some flexibility – for example, in the case of visual scene analysis POC, we use interchangeable camera lens (therefore enabling different fields of view) on top of a rather capable image sensor.

Going from a POC running adequate software to a final product design therefore requires some optimization efforts; nevertheless, GAPPoc is a good starting point to save time.

GAPPoc Variants

The first variant of GAPPoc, available today, targets Computer Vision in the visible spectrum.

Upcoming variants are in the works or under the definition, with target applications such as Scene Analysis in the near-IR and in the thermal IR spectrum, Sound Classification, Gesture Recognition, or Predictive Maintenance.

Incidentally, if you are a provider of sensors or other IoT-related devices that, like GAP8, bring decisive benefits at ultra-low power; GREENWAVES would like to hear from you and explore the opportunity to develop a proof of concept including your device.

First family member: GAPPoc-A, Computer Vision Concept Board

The first variant of GAPPoc (pictured in Fig. 1) focuses on low-power computer vision applications.

At its heart is a GAPMod core module integrating a GAP8 chip and a generously sized external memory, with 512Kbit of Flash and 64Mbit of RAM (complementing the internal 4Mbit RAM) on a Hyperbus delivering up to 1.2Mbit/s, to support even demanding classification algorithms.

It is complemented with a XVGA (752×480), 1/3’’ monochrome sensor (with binning/down-sampling options) with global shutter and optional wide dynamic range support – ON Semiconductor’s MT9V034. It comes with an S-mount into which an interchangeable standard M12 lens can be screwed. In a low power scenario, the sensor would typically be operated in snapshot mode and completely turned off between snapshots. The GAPPoc power supply generation scheme is designed to allow this. The objective is to obtain battery life greater than 1 year when taking on average one picture every 30 seconds at least.

The radio link is provided by a Bluetooth Low Energy (BLE) module from uBlox (NINA-B112, itself based on the nRF52832 SOC from Nordic), with a very compact on-board antenna. This module comes with pre-flashed firmware providing a serial link over BLE based on a proprietary BLE profile, as well as the possibility to define custom GATT profiles. This is controlled through AT commands over a UART interface. A third possible option is to completely rewrite Nina’s firmware, making use of Nordic’s SDK.

GAPPoc also offers an ultra-low profile Hirose DF40 connector bearing a selection of power supplies and interface signals (eg. I²C, SPI, GPIO, I²S…), which allows to securely clip a small satellite board to provide additional features. This could for example be a PIR, Time-of-Flight or ambient light sensors, microphones, etc.

An open-source software Board Support Package with example code is also provided.

Conclusion

With GREENWAVES GAP8 IoT Application Processor chip, you can analyze and classify all kinds of data just next to the sensors that produce them, at ultra-low power and without the privacy concerns of uploading this data to the cloud.

If you have in mind a product for which GAP8 would be a good fit and you would like to quickly come up with a Proof Of Concept representative of what your final product could offer, then the GAPPoc concept may help you.

The first member of the GAPPoc family targets understanding the contents of visual scenes. Other variants are being added, targetting applications such as scene analysis, gesture recognition, audio classification, or predictive maintenance.

GREENWAVES is keen to work with partners on the evolution of the GREENWAVES GAPPoc family, so please get in touch with GREENWAVES if you have an interest in this domain.

发送到邮箱 |
+1 赞 0
收藏
评论 0
| 转发至：

本文由雪飘梦飞转载自GREENWAVES Official Website，原文标题为:GAPPoc : A family of GAP8-centric Proof Of Concept boards for edge AI，本站所有转载文章系出于传递更多信息之目的，且明确注明来源，不希望被转载的媒体或个人可与我们联系，我们将立即进行删除处理。

全部评论（0）

暂无评论

GAP8 Performance Versus ARM M7 on Embedded CNNs

ARM recently published a new CMSIS library for embedded convolutional neural networks (CNNs) CMSIS-NN. Firstly, it was great to see ARM supporting the market that GreenWaves and GAP8 are focused on. We particularly liked their statement that: “Neural Networks are becoming increasingly popular in always-on IoT edge devices performing data analytics right at the source, reducing latency as well as energy consumption for data communication.”

产品发布时间 : 2024-09-10

Lynred and GreenWaves collaborate on New Occupancy Management Reference Platform for People Counting Sensor

GreenWaves and Lynred have collaborated on an open-source workspace management platform that allows quick deployment of sensors collecting accurate occupancy data. This platform combines Lynred‘s low-power IR sensors with GreenWaves‘ GAP8 processor to create battery-operated people counting devices, released under open source licenses. The platform ensures occupant anonymity using infrared technology and will be demonstrated at Embedded World in Nuremburg, Germany.

产品发布时间 : 2024-09-07

GreenWaves Technologies Partners with Open-Silicon to develop Industry’s First IoT Processor Based on PULP and RISC-V

Open-Silicon, a system-optimized ASIC solution provider, today announced it was selected by GreenWaves Technologies to develop GAP8, the industry’s first IoT processor. GAP8 is built on the open source Parallel Ultra Low Power (PULP) and RISC-V ISA projects. Open-Silicon is providing GreenWaves Technologies with the complete RTL-to-physical design custom SoC implementation that is required to transform this smart IoT concept into working silicon in volume production.

产品发布时间 : 2024-08-20

MELEXIS ACTUATORS-Melexis MLX81300 single-chip smart LIN Drivers for automotive motors

型号- MLX81208,MLX81109,MLX90380,MLX81301,MLX81200,MLX83203,MLX81300,MLX81206,MLX81107,MLX81207,MLX81108,MLX81325,MLX81326,MLX81205,MLX81106,MLX81150,MLX92212,MLX90290,MLX80300,MLX81310,MLX81311,MLX81316,MLX81315,MLX90B80,MLX81160,MLX83100,MLX90363,MLX902XX

商品及供应商介绍 - MELEXIS PDF 英文下载

2.4uA休眠功耗3*3mm封装霍尔位置传感器，16位分辨率数字输出，无磨损位置和角度检测

型号- MLX90316,MLX90317,MLX90393,MLX90363,MLX9039X,MLX90333,MLX9037X,MLX90324,MLX9036X

商品及供应商介绍 - MELEXIS - Apr 2021 PDF 中文下载

ROHM Developed A New Wi-SUN FAN Module with Radio Certification Can Connect up to 1,000 Devices

ROHM has developed a new Wi-SUN FAN (Field Area Network) module for infrastructure applications, capable of connecting up to 1,000 nodes in a mesh network.Wi-SUN FAN the latest Wi-SUN international wireless communication standard, which eliminates the communication costs associated with conventional LPWA (Low-Power Wide-Area) wireless standards. At the same time, Wi-SUN ensures superior reliability through multi-hop networks that automatically switches destinations after signal verification.

新产品发布时间 : 2021-03-08

Silicon Labs Launches Smallest, Lowest Jitter I²C-Programmable Crystal Oscillators

Silicon Labs has introduced a new family of small form-factor, high-performance crystal oscillators (XOs) and voltage-controlled crystal oscillators (VCXOs) for applications that require low jitter and frequency-flexible clock synthesis.The Si54x/6x Ultra Series XO/VCXOs deliver jitter performance as low as 80 femtoseconds (fs) for integer and fractional frequencies across the entire operating range.

新产品发布时间 : 2020-10-23

Chip vs. patch antenna for GNSS

型号- EB_NN02-224-1561-1606,NN02-224

应用笔记或设计指南 - IGNION - April 2021 PDF 英文下载

GAPMod 3.x GAP8 Centric Core Module with QSPI memories HARDWARE OVERVIEW

型号- GAPMOD3.0,GAP8,GAPMOD,GAPMOD 3.X

用户指南 - GREENWAVES - Rel. 1.2 - 11-Aug-21 PDF 英文下载

ROHM Develops Ultra-Low-Power On-Device Learning Edge AI Chip，Enabling Real-time Failure Prediction Without Cloud Server Required

ROHM has developed an on-device learning AI chip for edge computer endpoints in the IoT field. It utilizes artificial intelligence to predict failures (predictive failure detection) in electronic devices equipped with motors and sensors in real time with ultra-low power consumption.

新产品发布时间 : 2022-12-01

Amphenol Sensors(安费诺)AAS-AQS-UNO-RH-CO2空气质量传感器开发板产品亮点

型号- T9602,SM-PWM-01C,T6713,AAS-AQS-UNO-RH-CO2

用户指南 - AMPHENOL SENSORS - 03/2018 PDF 英文下载

超低功耗AI应用处理器

型号- GAP9,GAP10,GAP8

商品及供应商介绍 - GREENWAVES - 2021/4/23 PDF 中文下载

GAP8 HARDWARE INTEGRATION GUIDE

型号- GAP8

用户指南 - GREENWAVES - Rel.1.4 - Aug., 2020 PDF 英文下载

Silicon Labs（芯科科技）评价DONGLE 温度和湿度传感器Si7013USB-DONGLE 用户指南

型号- SI7013-A10-IM1,SI7013-A20-IM,SI7013USB-DONGLE,SI7013-A20-IM1R,SI7013-A20-YM1,SI7013-A10-GM,SI7013-A20-YM1R,SI7013,SI701X,SI7013-A20-GM1R,SI7013-A20-GM1,SI7013-A20-IM1,SI7013-A20-GMR,SI7013-A20-IMR,SI7013-A20-GM

用户指南 - SILICON LABS - 2017年03月07日 PDF 英文下载