Sipeed Maix Amigo AIoT Development Platform Based on RV64GC (RISC-V)

Features

• All-in-One Platform with Rich Peripherals
• RV64GC RISC-V 64-bit Dual-Core CPU for Powerful AI Applications
• Computing Power up to 1TOPS with dual cameras for Heavy Machine Vision Applications
• Built-in FPU, KPU, FFT Hardware Acceleration Units
• Built-in APU for High-Quality Audio Processing
• 3.5-inch capacitive touch screen for better user interaction
• Supports SPMOD and Grove modules to expand your projects

Description

SIPEED Maix Amigo is an all-in-one programmable AIoT development kit that can be used for AI and IoT learning.

SIPEED Maix Amigo has a built-in RV64GC RISC-V 64-bit 400Mhz dual-core high-performance processor with 8M on-chip SRAM and 16MiB FLASH storage. It is equipped with 2 cameras, TF card slot, user buttons, TFT capacitive touch screen, lithium battery, speaker, microphone, 2 USB ports, and an expansion interface.

It has a variety of hardware acceleration units such as KPU, FPU, FFT, etc. and the total computing power can be up to 1TOPS, which can conveniently realize machine vision/hearing algorithms of various application scenarios, and can also perform pre-processing work of voice scanning and voice data output. Users can use Maix Amigo to easily build a face recognition access control system.

The hardware integrates front and rear 0.3-megapixel cameras and a 3.5-inch capacitive touch screen.

A Wide Range of Sipeed Modules (SP-MOD) and Grove Modules Supported

Maix Amigo can connect to a variety of peripheral modules and sensors through external 3 Grove interfaces and 3 SP-MOD interfaces to meet your multiple application needs.

What is Grove?

Grove makes it easier to connect, experiment, and simplify the prototyping process. No jumpers or soldering required. We have developed more than 300 Grove modules, covering a wide range of applications that can fulfill a variety of needs. Not only are these open hardware, but we also have open-source software. You can easily connect Grove modules to the Maix Amigo to further expand your projects!

What is MaixPy?

The software for Maix Amigo is equipped with MaixPy by default, and users can easily use MicroPython syntax to quickly get started with AloT development, and develop rich Al applications such as image recognition, face recognition, object recognition, and object classification.

MaixPy is a project of porting Micropython to RV64GC chip that is, users can finally control the function of RV64GC chip through Micropython programming. For example, you can directly call the facial recognition algorithm built into the firmware through Micropython programming, and finally generate a Micropython file, which can be downloaded to the Flash chip, and run on it. In addition, MaixPy supports the normal operation of MCU and integrates machine vision and microphone arrays to rapidly develop intelligent applications in the field of AIoT with extremely low cost and practicality.

What is the Development Environment on Maix Amigo?

By default, the development board has burned MaixPy firmware (the firmware source code is open source and can be downloaded from Github). MaixPy has developed a variety of functional libraries that developers can call directly. Maix series products use Maix IDE developed by the Sipeed team. The software is free of installation. After downloading, you can directly use MaixPy to use Micropython script syntax. Developers can edit the script on the computer and upload it to the development board to execute the script directly on the development board. The IDE can view camera images in real-time and save files to the development board, which is very convenient for development.

In addition to MaixPyIDE, developers can also use ArduinoIDE, PIatformIO IDE, and other environments for development, please refer to our technical information for details.

Specifications

Specification	Description
CPU	RV64GC RISC-V Dual-Core 64-bit with FPU
CPU Frequency	400MHz(can be overclocked to 500MHz)
SRAM	Built-in 8MiB
Storage	16MiB Flash, support Micro-SD (up to 128GB)
Display	3.5” TFT Capacitive Touch Screen, 320*480
Camera	Front Camera: GC0328, 0.3M, 30fps Rear Camera: OV7740, 0.3M, 60fps
Speech recognition	Offline speech recognition, sound field imaging (6+1mic)
Network model	Support MobileNetV1/V2, TinyYOLOv2, face recognition, ASR, etc.
Deep learning framework	Support TensorFlow lite (recommended), ONNX, paddlepaddle model conversion
Peripherals	FPIOA, UART, GPIO, SPI, 12C, 12S, TIMER
Hardware acceleration unit	KPU (convolution operation accelerator) FPU (Floating Point Accelerator) APU (Audio Processor) FFT (Fourier transform accelerator)
Onboard resources	3.5 inch TFT screen Reset button * 1 Function button * 3 Front and rear camera * 2 LED * 3 Acceleration sensor * 1
Onboard interface	USB Type-C interface Audio interface (support external speakers) TF card slot Grove standard interface SP-MOD interface (support SP-MOD interface module)
Dimensions	104.3*63.3*16.5mm
Power supply	USB-type C or internal lithium battery (520mAh)

Software Development

• Software environment: MaixPy (MicroPython)
• Development environment: MaixPy IDE, PlatformlO IDE, Arduino IDE, etc.
• Programming language: C, MicroPython

Applications

• Image Classification
• Object Detection
• Face Recognition
• Speech Recognition
• Game Simulation

Hardware Overview

 

1	Front camera	9	USB-C port (OTG)
2	Touch screen	10	2 x SPMOD interface
3	USB-C port (power and debug)	11	Selection key
4	3 x Grove ports	12	Enter key
5	Rear camera	13	TF card slot
6	Flash	14	Reset button
7	Indicator LEDs	15	Power button
8	SPMOD interface

 

Part List

1 x Sipeed Maix Amigo