LC-AI-K210

From LCWIKI
Jump to: navigation, search

Product name

KI210 Machine Vision.


Overview

The Elsey Dock version K210 development board is equipped with the Sipeed M1/M1W AI module. The core of the module is a system-on-chip (Soc) Kendryte K210 that integrates machine vision and machine hearing capabilities.Using TSMC's ultra-low power consumption 28nm advanced process, with a dual-core 64-bit processor, it has better power consumption performance, stability and reliability.The solution strives for zero-threshold development, and can be deployed in the user's product in the shortest time, endowing the product with artificial intelligence. Kendryte K210 is positioned as a SoC for the AI ​​and IoT market,At the same time, it is a very convenient MCU. The Chinese meaning of Kendryte is Kanzhi, and Kanzhi comes from Kanwu Tanzhi. The main application field of this chip is the field of Internet of Things, and it is developed in the field of Internet of Things, so it is a survey object;This chip mainly provides artificial intelligence solutions and explores in the field of artificial intelligence, so it is called Tanzhi.


Function

  • Onboard Sipeed M1 or M1W module, containing Kendryte K210 artificial intelligence Soc chip
  • CPU: RISC-V 64bit dual-core processor, 400Mhz standard frequency (overclockable)
  • FPU specification: meet IEEE754-2008 standard
  • Image recognition: QVGA@60FPS/VGA@30FPS
  • Deep learning framework: TensorFlow/Keras/Darknet
  • Peripherals: FPIOA, UART, GPIO, SPI, I²C, I²S, WDT, TIMER, RTC etc
  • Neural Network Processor (KPU):

1.Support fixed-point models trained by mainstream training frameworks according to specific restriction rules
2.There is no direct limit to the number of network layers, and each layer of convolutional neural network parameters is supported to be configured separately, including the number of input and output channels, input and output row width and column height
3.Support two convolution kernels 1x1 and 3x3
4.Support any form of activation function
5.The maximum supported neural network parameter size is 5.5MB to 5.9MB when working in real time
6.The maximum supported network parameter size when working in non-real time is (Flash capacity - software volume)

  • Static Random Access Memory (SRAM): SRAM consists of two parts, 6MB of on-chip general-purpose SRAM memory and 2MB of on-chip AI SRAM memory, for a total of 8MB
  • Field Programmable IO Array (FPIOA/IOMUX): FPIOA allows users to map 255 internal functions to 48 free IOs on the periphery of the chip
  • Fast Fourier Transform Accelerator: The FFT accelerator implements FFT operations in hardware
  • Software overview:

1.FreeRtos & Standard SDK: Support FreeRtos and Standrad development kit
2.MicroPython Support: Support MicroPython on M1
3.Machine vision: Machine vision based on convolutional neural network
4.Machine hearing: High performance microphone array processor

  • Onboard MSM261S4030H0 MEMS silicon microphone
  • DVP camera interface: 24P 0.5mm FPC connector
  • LCD interface: 24P 0.5mm FPC connector (8bit MCU LCD)
  • Onboard TM8211 I2S audio chip and PAM8403 power amplifier IC (maximum support 2X3W output)
  • Onboard Micro SD card slot
  • Automatic download circuit: directly use the USB Type-C cable to download the program
  • All I/O ports are led out through 2.54mm pin headers
  • Onboard Boot button and Reset button
  • Wireless WiFi (optional): only the development board with Sipeed M1W module has WiFi function, ESP8285 chip, supports 2.4G 802.1.1.b/g/n standard
  • Development board accessories: OV2640 camera module, 2.4-inch 240*320 resolution ST7789 driver chip 8-bit parallel LCD, Micro USB to Type-C head, 2.54mm IO port lead-out pin header
  • External power supply voltage requirement: 5.0V ±0.2V
  • External power supply current requirement: >300mA @ 5V



Introduction

1.Board size:60*45mm Weight:27g(Including accessories)
K210-1.png

2.Introduction for onboard resources
K210-2.png

K210-3.png
(1)Sipeed M1/M1W module: Contains Kendryte K210 artificial intelligence chip
(2)USB Type-C interface: can be used for board power supply/program download/firmware upgrade, etc
(3)500mA patch self-recovery fuse: avoid overloading the computer USB port
(4)RESET button
(5)BOOT key
(6)Programmable 3528RGB: use IO0--IO2 control
(7)Audio output: L+ and L- are left channel, R+ and R- are right channel, maximum support 2X3W output
(8)PAM8403 power amplifier IC
(9)CH340C chip: USB to TTL serial port IC, can be used for program download/firmware upgrade, etc
(10)DVP camera interface: 24P 0.5mm FPC connector, connected to OV2640 camera by default
(11)LCD interface: 24P 0.5mm FPC connector, connected to 2.4-inch 240*320 resolution ST7789 driver chip 8-bit parallel port LCD by default
(12)Micro SD card (TF card) slot
(13)TM8211 I2S DAC audio chip
(14)MSM261S4030H0 silicon microphone: MEMS microphone with omnidirectional, top port I2S output, high performance and high reliability



Hardware connection

1.Install MaixPy IDE
The K210 Dock development board uses MaixPy IDE as the integrated development environment under Windows. Double-click maixpy-ide-windows-0.2.3.exe to install the IDE. ” button to run the program.
K210-4-1.jpg


2.Install the CH340 serial port driver
The K210 Dock development board communicates with the host computer through the CH340 USB to UART chip, such as: downloading programs/updating firmware, so you need to install the driver first, double-click CH341SER.EXE, after the driver is successfully installed, you will see the port number in the computer device manager. as follows:
K210-5.png K210-6.png


3.Development board basic test
(1) start up
●Test items: Type-C USB, K210 module, LCD.
●Test method: Connect the development board to the Type-C USB cable, and the default boot interface will be displayed, as follows:
K210-7.png

(2) RGB flashing light
●Test items: Type-C USB, CH340C chip, K210 module, 3528 RGB lights.
●Test method: Connect the development board to the computer through the Type-C USB cable, use MaixPy IDE to open the program "RGB.py", click "Connect" at the bottom left of MaixPy IDE, click "Start" after the connection is successful, and the RGB light will be red- -Green--blue cycle flashing, as follows:
K210-8.png

K210-9.png

(3)Camera
●Test items: Type-C USB, CH340C chip, K210 module, OV2640 camera, LCD.
●Test method: Connect the development board to the computer, connect the camera and LCD, run the program "Camera.py", the IDE and LCD will display the images captured by the camera in real time, as follows:
K210-10.png

K210-11.png

(4)Play video
●Test items: Type-C USB, CH340C chip, K210 module, SD card, LCD, PT8211 I2S audio chip, PAM8403 power amplifier chip.
●Test method:
①Copy the test video "badapple.avi" to the SD card
K210-12.png
Note: If you want to play other AVI format videos, please make sure that the video size is 320x240, the video compression format is mjpeg, the audio compression format is PCM, and the video path in the program should be modified accordingly.

②Connect the development board to the computer, connect the LCD, insert the SD card, connect the amplifier output terminal L+L-/R+R- to the speaker, and run the program "Play_Avi_Video.py" to play the video, as follows:
K210-13.png

K210-14.png
For more instructions, please move to:https://maixpy.sipeed.com/zh/

Retrieved from "http://wiki.lctech.cc/index.php?title=LC-AI-K210&oldid=393"