Sipeed MAIX: AI at the edge

AI is pervasive today, from consumer to enterprise applications.

With the explosive growth of connected devices, combined with a demand for privacy/confidentiality, low latency and bandwidth constraints, AI models trained in the cloud increasingly need to be run at the edge.

MAIX is Sipeed’s purpose-built module designed to run AI at the edge, we called it AIoT.

It delivers high performance in a small physical and power footprint, enabling the deployment of high-accuracy AI at the edge, and $6's price make it possible embed to any IoT devices.

As you see, Sipeed MAIX is quite like Google edge TPU, but it act as master controller, not an accelerator like edge TPU, so it is more low cost and low power than AP+edge TPU solution.

MAIX's Advantage and Usage Scenarios:

MAIX is not only hardware, but also provide an end-to-end, hardware + software infrastructure for facilitating the deployment of customers' AI-based solutions.

Thanks to its performance, small footprint, low power, and low cost, MAIX enables the broad deployment of high-quality AI at the edge.

MAIX isn't just a hardware solution, it combines custom hardware, open software, and state-of-the-art AI algorithms to provide high-quality, easy to deploy AI solutions for the edge.

MAIX can be used for a growing number of industrial use-cases such as predictive maintenance, anomaly detection, machine vision, robotics, voice recognition, and many more. It can be used in manufacturing, on-premise, healthcare, retail, smart spaces, transportation, etc.

Catalog :

• MAIX's CPU
• MAIX's Module
• MAIX's Development Board
• MAIX's Peripheral Module
• MAIX's Software
• MAIX's Deep learning 
• MAIX's Community
• Support options & Developers plan

MAIX's CPU

In hardware, MAIX have powerful KPU K210 inside, it offers many excited features:

1. First competitive RISC-V chip, also Fisrt competitive AI chip, newly release in Sep. 2018.
2. It is 28nm process, dual-core RISC-V 64bit IMAFDC,  on-chip huge 8MB high-speed SRAM(not for XMR :D), 400MHz frequency(able to 800MHz).
3. It have KPU(Neural Network Processor) inside,  64 KPU which is 576bit width, support convolution kernels, any form of activation function. It offers 0.25TOPS@0.3W,400MHz, and when you overclock to 800MHz, it offers 0.5TOPS. It means you can do object recognition 60fps@VGA.
4. It have APU(Audio Processor),  support 8mics, upto 192KHz sample rate, hardcore FFT unit inside, thus it is easy to make an Mic Array(MAIX offer it too)
5. It have flexible FPIOA(Field Programmable IO Array), you can map 255 functions to all 48 GPIOs on the chip.
6. DVP camera and MCU LCD interface, you can connect an DVP camera, run your algorithm, and display on LCD.
7. And many other accelerators and peripherals: AES Accelerator, SHA256 Accelerator, FFT Accelerator(not APU's one), OTP, UART,WDT,IIC,SPI,I2S,TIMER,RTC,PWM, etc.

MAIX's Module

Inherit the advantage of K210's small footprint, Sipeed MAIX-I module, or called M1, integrate K210, 3ch dcdc power, 8MB/16MB/128MB Flash (M1w module add wifi chip esp8285 on it) into Square Inch Module.  All usable IO breaks out as 1.27mm(50mil) pins, and pin's voltage is selectable from 3.3V and 1.8V. 

In addition, as many DIYer want build their own work with breadboard, we newly provide breadboard-friendly board for you, it called MAIX-Micro, or called MM.

It is twice of M1 size, 1x2 inch size, breadboard-friendly, and also SMT-able,

It integrate USB2UART chip, auto download circuit, RGB LED, DVP Camera FPC connector(support small FPC camera and standard M12 camera), MCU LCD FPC connector(support our 2.4 inch QVGA LCD), TF card solt.

BTW, MM is able to adjust core voltage! you can adjust from 0.8V~1.2V, overclock to 800MHz!

We are looking forward to popularize AIoT, so we reduce MM's Flash size to 8MB, and cost down MM to 5$ in this crowdfunding, Cheers!

Also, we have PLCC and NGFF module in development, if you have any requirement, you can contact us via support@sipeed.com

MAIX's development board

Firstly, We make an prototype development board for M1, called M! dock or Dan Dock, it is simple, small, cheap, but all functions include.

It is in mass production already, it will send in one week if you support this suit.

Here have an simple demo to detect face:

Then, we make bigger and better one:

It is 88x60mm, all pins out, with standard M12 lens DVP camera, and the Camera can be fliped  from front to rear!

It have on board JTAG&UART based on STM32F103C8, so you can debug M1 without extra Jlink.

It have lithium battery manager chip with power path management function, you can use the board with lithium battery and usb power without conflict~

It have  I2S Mic, Speaker, RGB LED, Mic array connector, thumbwheel, TF card Slot and so on.

This suit include 2.8 inch LCD too, and have an simple case for it. 

MAIX's peripheral module

We have see basic camera and LCD interface on board, so DVP Camera and MCU LCD is support very early.

It support OV7725， OV2640(default, 2M)，OV5640 for the moment.

And it support 2.4 inch st7789, and 2.8 inch ili9341 LCD in QVGA resolution.

For MM module, we have extra tiny I2S mic module:

And We have Cool  6+1 Microphone Array Module!

here is the demo of microphone array

MAIX's SoftWare

MAIX support original standalone SDK, FreeRTOS SDK base on C/C++.

And we port micropython on it: http://en.maixpy.sipeed.com/

It support FPIOA,  GPIO, TIMER, PWM,  Flash, ov2640， LCD, etc.

And it have zmodem, vi, SPIFFS on it, you can edit python directly or sz/rz file to board.

We are glad to see you contribute for it:

https://github.com/sipeed/MaixPy                 //Maixpy project

https://github.com/sipeed/MaixPy_Doc_Us_En_Backup           //Maixpy wiki project

MAIX's Deep learning

MAIX support fixed-point model that the mainstream training framework trains
according to specific restriction rules, and have model compiler to compile models to its own model format.

It  support tiny-yolo, mobilenet-v1, and , TensorFlow Lite !

Many TensorFlow Lite model can be compiled and run on MAIX~

And We will soon release model shop, you can trade your model on it.

MAIX's community

our wiki:  en.maixpy.sipeed.com,  en.dan.sipeed.com

our bbs:  http://bbs.lichee.pro/t/sipeed-m1

our telegram group:  https://t.me/sipeed

our qq group(3K member):  826307240

our qq group for AI purpose:  878189804

our FAE support email:  support@sipeed.com

Support options & Developers plan

MAIX-I module:  (in stock)

$55    MAIX-I module x 10

$65    MAIX-I wifi version module x10

MAIX Bit:(December)

$5    Early Bird! MAIX Bit for 5$   

$6    MAIX Bit for 6$

$15    MAIX Bit Suit With LCD, Camera

$30    MAIX Bit x 5

M1/M1w dock  (in stock)

$15    Goods in stock! M1 dock suit  (M1 dock(mic and pa on board ) + 2.4 inch LCD + ov2640)

$16    Good in stock! M1w dock suit  (M1w (w means wifi) dock(mic and pa on board ) + 2.4 inch LCD + ov2640)

$75    M1 dock suit x5

$80    M1w dock suit x 5

MAIX GO (December)

$22    MAIX GO Suit    (MAIX GO (WiFi，debugger,mic, speaker, battery chip,  acceleration sensor on board ) + 2.8 inch LCD + ov2640 with M12 lens)

$45    ALL IN ONE: MAIX GO Suit  (MAIX GO Suit + R6+1 Microphone Array + Binocular camera)

Accessories

$7    I2S MIC x 4 (November )  

$12    MAIX R6+1 Microphone Array   (November)

$12    MAIX Binocular Camera (November or December)

Others

$20    DHL express fill post
$99   Consulting fees

Team Introduction

Sipeed is an young team from Shenzhen, China.

We are expert at  IoT Application, Embeded System, Deep Learning.

We aim to popularize AIoT,  promoting AI progress, make our life better.

K210 is an epochal AI Chip ,  so we choose it and build the MAIX project to pushing it  all over the world.

Let's Sipeed up, Maximize AI's power!