Last year we wrote about Kendryte K210 dual core RISC-V processor specifically designed for for machine vision and machine hearing as well as the corresponding Kendryte KD233 which enables inference at the edge, e.g. tasks such as face recognition or object detection.
Latter on we found the processor in Sipeed M1 module which went for as low as $5 in a crowdfunding campaign, and was fitted to some low cost boards now selling for $12.90 on Seeed Studio. The latter company has now designed Grove AI HAT that aims to assist Raspberry Pi in running the edge computing workloads previously described, as exposes 6 Grove interfaces to extend functionality with some of the Grove add-on modules.
Grove AI HAT specifications:
- AI Module – Sipeed “MAIX” M1 with Kendryte K210 dual core RISC-V processor @ 600 MHz, KPU Convolutional Neural Network (CNN) hardware accelerator, APU audio hardware accelerator, 8 MB general purpose SRAM including 5.9MB usable as AI SRAM memory
- Camera I/F – DVP camera support; two mounting locations (top and bottom of the board)
- Display I/F – Supports for Sipeed 2.4″ QVGA LCD display
- Audio – Built-in microphone
- USB – 1x USB 2.0 Device type C for power and programming
- 40-pin Raspberry Pi connector with I2C/UART/SPI/I2S/PWM/GPIO.
- 6x Grove connectors: 2x analog, 1x UART, 1x I2C, 1x PWM, 1x digital
- Debugging – 8-pin UART/JTAG header
- Misc – Boot & reset buttons, boot (user programmable) & power LEDs, switch for RPi 5V, on-board 3-axis accelerometer
- Power Supply – 5V via USB type-C port, or Vin pin on Raspberry Pi header
- Dimensions – 83 x 55 mm (4x Raspberry Pi HAT mounting holes)
The board can either be used in conjunction with a Raspberry Pi board, or in standalone via a USB-C cable. It supports the baremetal Kendryte Standalone SDK, as well as ArduinoCore K210 for the Arduino IDE working in Linux, Windows, and Mac OS meaning it’s possible to leverage the Grove Arduino libraries and other Arduino libraries on this board.
Pre-order price is $24.50 after which it will be $28.50, and shipping is scheduled for mid-June. Seeed Studio is also planning to release a kit with the camera and an LCD as found in the company’s Sipeed MAIX BiT kit.
I have not seen video demos on Grove AI Hat, but there are a few on Caesar Wu YouTube channel running on Kendryte K210 hardware platform, including the tiny-yolov2 demo embedded below.
Jean-Luc started CNX Software in 2010 as a part-time endeavor, before quitting his job as a software engineering manager, and starting to write daily news, and reviews full time later in 2011.
One Reply to “Grove AI HAT Helps Raspberry Pi Run Edge Computing Workloads”
Thats actually really cool. And i have a Sipeed M1 but didnt knew about the ArduinoCore K210, which should help me a kot since a dont really liked working with the mirocpython on the board