Radxa Dragon Q6A User Guide

This document describes how to use the Aidge AI toolchain. If you encounter issues with hardware interfaces or the underlying system, please refer to the Radxa Dragon Q6A Official Documentation.

1. Quick Start

1.1 Environment Setup

1.1.1 Hardware Requirements

• Mainboard: Radxa Dragon Q6A
• System boot medium: MicroSD card
• Power adapter: 12V Type-C power adapter (PD protocol compatible)
• USB-to-serial cable (optional): For serial debugging and system login
• HDMI cable and monitor (optional): For displaying the system interface and GUI operations

1.2 System Installation and Usage

Please refer to the system installation chapter in the Radxa Dragon Q6A official documentation for downloading, installing, and using the system. Documentation link: Radxa Dragon Q6A

2. Aidge AI Toolchain Usage

2.1 Install the Toolchain

• Configure the Aidge apt repository

sudo wget -qO- https://archive.aidlux.com/ubuntu24/public.key | sudo gpg --dearmor -o /etc/apt/trusted.gpg.d/private-aidlux.gpg

echo 'deb [arch=arm64 signed-by=/etc/apt/trusted.gpg.d/private-aidlux.gpg] https://archive.aidlux.com/ubuntu24 noble main' | sudo tee /etc/apt/sources.list.d/private-aidlux.list

sudo apt update

• Install AidLite and license

sudo apt install -y aidlite-sdk aidlite-qnn236 aidlite-qnn240 aidlite-qnn248 aid-lms aidlms-sdk

• Install dependencies for compiling C++ examples

sudo apt install -y build-essential cmake g++ libopencv-dev

2.2 Aidlite SDK Usage Examples

• SDK & API Documentation

AidLite SDK

2.2.1 aidlite_qnn236

# Navigate to the project directory:
cd /usr/local/share/aidlite/examples/aidlite_qnn236/cpp

# Compile the executable:
sudo mkdir build && cd build
sudo cmake ..
sudo make

# Run the test project script:
sudo ./qnn_yolov5_multi 3
# Note: 1. sudo requires password: radxa  2. This example only invokes DSP

2.2.2 aidlite_qnn240

# Navigate to the project directory:
cd /usr/local/share/aidlite/examples/aidlite_qnn240/cpp

# Compile the executable:
sudo mkdir build && cd build
sudo cmake ..
sudo make

# Run the test project script:
sudo ./qnn_yolov5_multi 3
# Note: 1. sudo requires password: radxa  2. This example only invokes DSP

2.2.3 aidlite_qnn248

# Navigate to the project directory:
cd /usr/local/share/aidlite/examples/aidlite_qnn248/cpp

# Compile the executable:
sudo mkdir build && cd build
sudo cmake ..
sudo make

# Run the test project script:
sudo ./qnn_yolov5_multi 3
# Note: 1. sudo requires password: radxa  2. This example only invokes DSP

3. Licensing

The AI toolchain has a 30-day trial period. If you need a license, please contact Aidge's FAE to obtain a license code. The steps are as follows:

• Retrieve the device serial number

cat  /sys/devices/soc0/serial_number

• Send the device serial number to Aidge's FAE to obtain a license code