1. Introduction
This manual provides essential instructions for the proper installation, operation, and maintenance of your Yahboom Raspberry Pi AI HAT+ 8L-13TOP Artificial Intelligence Hailo-8L Accelerator. This HAT (Hardware Attached on Top) is designed to enhance the AI capabilities of your Raspberry Pi 5, offering significant AI acceleration for various edge computing applications.
2. Product Overview
2.1 Key Features
- AI Computing Power: Provides 13 TOPS (Tera Operations Per Second) for enhanced AI acceleration on Raspberry Pi 5.
- Framework Compatibility: Supports common AI frameworks such as TensorFlow, PyTorch, TensorFlow Lite, Keras, and ONNX.
- Camera Software Integration: Fully integrates with the Raspberry Pi's camera software stack for post-processing tasks like object detection, image segmentation, and pose estimation.
- Adaptability: Designed for stacking installation with Raspberry Pi 5 and supports active heat sinks for optimal thermal management.
- Edge AI Performance: Enables real-time, low-latency, and efficient AI inference on edge devices.
- Standard M.2 Interface: Facilitates quick integration.
- Wide Temperature Range: Supports an extended operating temperature range of -40°C to 85°C (Industrial) and -40°C to 105°C (Automotive).
2.2 What's in the Box
- Yahboom Raspberry Pi AI HAT+ (8L-13TOP)
- Hailo power module
- Pin header
- PCIe cable
- Copper column screws
3. Setup Instructions
3.1 Hardware Installation
- Prepare Raspberry Pi 5: Ensure your Raspberry Pi 5 is powered off and disconnected from all peripherals.
- Attach AI HAT+: Carefully align the AI HAT+ with the GPIO pins and PCIe connector on your Raspberry Pi 5. Gently press down to ensure a secure connection.
- Connect PCIe Cable: Use the provided PCIe cable to connect the AI HAT+ to the Raspberry Pi 5's PCIe interface.
- Secure with Screws: Use the copper column screws to firmly attach the AI HAT+ to the Raspberry Pi 5, providing stability.
- Install Heat Sink (Recommended): For optimal performance and to prevent overheating during high-load AI computations, it is highly recommended to install an active heat sink on the Raspberry Pi 5 and potentially on the AI HAT+ if applicable. This ensures the AI acceleration module is adequately cooled.
Image 1: The Yahboom Raspberry Pi AI HAT+ 8L-13TOP Accelerator board, showing the Hailo chip and connectors.
Image 2: The AI HAT+ installed on a Raspberry Pi 5, connected to a monitor and keyboard, demonstrating a typical setup.
Image 3: Exploded view illustrating the installation of the AI HAT+ expansion board, PCIe cable, official active radiator, pin header, and Raspberry Pi 5. Note: Raspberry Pi 5 and active radiator are not included.
3.2 Software Configuration
The Raspberry Pi AI HAT+ is designed to integrate seamlessly with the Raspberry Pi's camera software stack. For detailed software setup, driver installation, and framework configuration, please refer to the official Yahboom technical support resources and documentation. This typically involves installing necessary libraries and tools to utilize the Hailo-8L accelerator with your chosen AI frameworks.
Image 4: Diagram illustrating the Hailo comprehensive software tool chain and developer tools, including building environment, model build computer, runtime environment, and real-time runtime library.
4. Operating Instructions
4.1 AI Acceleration
Once the hardware is installed and software configured, the AI HAT+ will accelerate AI inference tasks. You can deploy various AI-driven applications for process control, home automation, research, and more, leveraging the 13 TOPS computing power of the Hailo-8L chip.
4.2 Camera Integration
The AI HAT+ enhances the Raspberry Pi's camera capabilities by offloading neural network processing to the accelerator. This allows for efficient execution of tasks such as object detection, image segmentation, and pose estimation directly from the camera feed.
Image 5: The AI HAT+ installed on a Raspberry Pi 5, with a camera module connected via a ribbon cable, demonstrating its use in vision-based AI applications.
Image 6: Visual comparison of actual measurement results for AI tasks like YOLOV5 object detection, pose estimation, and background segmentation, showcasing the performance of the Hailo-8L acceleration module.
5. Maintenance
- Keep Clean: Regularly clean the HAT and Raspberry Pi to prevent dust accumulation, which can affect performance and cooling. Use compressed air or a soft brush.
- Thermal Management: Ensure any installed heat sinks or cooling fans are functioning correctly. Overheating can lead to reduced performance and potential damage.
- Firmware Updates: Periodically check for firmware and software updates from Yahboom or Hailo to ensure optimal performance and access to new features.
- Handle with Care: Avoid static discharge when handling the board. Store in an anti-static bag when not in use.
6. Troubleshooting
- HAT Not Detected: Verify that the HAT is correctly seated on the GPIO pins and the PCIe cable is securely connected. Check software drivers and configuration.
- Performance Issues: Ensure adequate cooling is provided, especially during intensive AI tasks. Check for background processes consuming resources. Verify software configuration and model optimization.
- System Instability: Ensure your Raspberry Pi 5 power supply is sufficient. An underpowered system can lead to instability.
- Camera Integration Problems: Confirm the camera module is correctly connected and enabled in the Raspberry Pi OS. Check for compatibility issues with specific camera models and software versions.
For further assistance, please refer to the technical support section.
7. Specifications
| Feature | Specification |
|---|---|
| Model Name | Raspberry Pi AI HAT+ |
| Item Model Number | 8L-13TOP AI HAT+ |
| Brand | Yahboom |
| AI Computing Power | 13 TOPS (INT8) |
| Supported Frameworks | TensorFlow, TensorFlow Lite, ONNX, Keras, PyTorch |
| Supported Operating Systems | Raspberry Pi OS (Latest version), Linux |
| Software Support | Fully integrated into Raspberry Pi's camera software stack |
| Supported Host Architectures | Raspberry Pi 5 |
| Working Temperature | 0°C-50°C (Industrial: -40°C to 85°C, Automotive: -40°C to 105°C for Hailo-8L chip) |
| Physical Dimensions | 56.5mm (W) × 65mm (L) × 5.5mm (H) |
| PCIe Interface | Compatible with Pi5 PCIe Gen3 standard version |
| HAT Interface | Conforms to Raspberry Pi HAT+ specification |
| Included Components | Hailo power module |
Image 7: Detailed parameter comparison table for Hailo-8 AI HAT+ (26 TOPS) and Hailo-8L AI HAT+ (13 TOPS), outlining performance, supported frameworks, operating systems, software, host architectures, working temperature, physical dimensions, PCIe interface, and HAT interface.
8. Warranty and Support
Yahboom provides high-quality after-sales technical support services for the Raspberry Pi AI HAT+. If you encounter any problems during use, please contact Yahboom customer service or technical support for assistance. Please refer to your purchase documentation for specific warranty terms and conditions.
Technical Support Link: http://www.yahboom.net/study/hailo8
Image 8: Overview of case studies demonstrating object detection, posture estimation, and image segmentation using the rpicam-apps camera application, alongside a section detailing available learning materials and tutorials.
