Raspberry Pi unveils the new Raspberry Pi AI HAT+, building on their successful relationship with Hailo. The Raspberry Pi AI HAT+ is an add-on board that includes a Hailo AI accelerator for the Raspberry Pi 5. It provides an accessible, low-cost, and energy-efficient alternative for incorporating high-performance AI into your applications. Application areas include process control, security, home automation, and robotics. This certified product unlocks the immense potential of AI, especially when teamed with the Raspberry Pi 5.
The AI HAT+ is available in 13 tera-operations per second (TOPS) configurations, based on the Hailo-8L and Hailo-8 neural network inference accelerators. The 13 TOPS variants can execute neural networks for applications like as object detection, semantic and instance segmentation, pose estimation, and more. The 13 TOPS variation can handle larger networks, run them faster, and run several networks more efficiently.
Hailo accelerators go beyond pure power. It promises to be the market’s most efficient edge AI processor, outperforming competitors in terms of both area and power usage by orders of magnitude. This efficiency is critical when executing deep learning models on resource-constrained, battery-powered edge devices such as the Raspberry Pi 5. Furthermore, the Hailo accelerators use a unique architecture optimized for neural networks known as the Neural Processing Unit (NPU). It operates over a wide temperature range and has the scalability to handle many data streams and models at the same time, making it an adaptable option for a variety of AI applications.
The AI HAT+ communicates via the Raspberry Pi 5’s PCIe Gen 3 interface. When the host Raspberry Pi 5 runs the latest Raspberry Pi OS image, it instantly recognizes the onboard Hailo accelerator and makes the NPU available for AI computing tasks. It also works flawlessly with the Raspberry Pi OS and the current camera software stack (libcamera, rpicam-apps, and picamera2). This allows you to use existing libraries and tools within your familiar development environment, saving time and expediting the creation process. Of course, using the apt package manager, you can easily install the driver and software package (HailoRT, Hailo Tappas, and more) for the Hailo accelerators from the existing Raspberry Pi OS.
The HAILO AI Accelerator offers:
- Performance: Up to 26 TOPS (Tera Operations Per Second) processing capability, making it efficient for real-time AI tasks.
- Efficiency: Hailo claims to outperform all other edge AI processors in area and power efficiency by an order of magnitude.
- Architecture: Utilizes a structure-driven dataflow architecture to optimize neural networks.
- Power Consumption: Low power consumption allows for running deep neural network inference on edge devices.
- Temperature Range: Supports a wide temperature range of 0°C to 50°C for various environments.
- Scalability: Enables simultaneous processing of multiple data streams and models, making it suitable for complex AI applications.
Perfect for Computer Vision and Edge AI Robotic
BTW, the AI HAT+ is fully compatible with the latest Raspberry Pi OS (Bookworm) and also all the camera modules and the software stack. We are very excited about this innovation from Raspberry Pi, it empowers many of us to:
- Develop cutting-edge AI applications:
- Image recognition and classification
- Object detection and tracking
- Facial recognition
- Natural Language Processing (NLP) tasks (with additional software)
- And much more!
- Elevate your projects:
- Add intelligent features to your Raspberry Pi creations
- Experiment with real-world AI applications
- Push the boundaries of creativity and innovation
You will see a leap improvement in existing computer vision projects based on Raspberry Pi, for example, the Donkeycar, and OpenCV image processing projects. The Raspberry Pi AI HAT+ for Raspberry Pi 5 is the perfect solution for developers, hobbyists, and educators seeking to explore the exciting world of AI. Join the revolution and take your Raspberry Pi projects to the next level!
SPECIFICATIONS
- Hailo-8 or Hailo-8L accelerator offering 13 TOPS inferencing performances respectively.
- FPC/FFC length is 30mm and preassembled on the HAT too.
- Supports single-lane PCIe 2.0 interface (500MB/s peak transfer rate)
- Conforms to the Raspberry Pi HAT+ Specification
- Supports installation of the official active cooler.
- Minimalist design, easy to install
- The HAT+ comes with an LED indicator for power
- HAILO AI Accelerator:
- HAILO 8L: Offer 13 Tera (Trillion) Operations Per Second (TOPS) AI Processing
- Interface: PCIe Gen 3.0, 2-lanes
- Supported host architectures: X86 or ARM
- Supported OS: Linux, Windows. In this case, Raspberry Pi OS is fully support
- Supported AI frameworks: TensorFlow, TensorFlow Lite, Keras, PyTorch & ONNX
- Best-in-class power efficiency
- Operating Temperature: 0℃ to 50℃ (ambient)
- Work out of the box with Raspberry Pi 5 and Raspberry Pi OS
- Fully integrated with Raspberry Pi Camera Software Stack, ready for computer vision development and deployment.
- Dimension: 65.0mm x 56.5mm x 5.6mm (max)
SHIPPING LIST
- 1 x AI HAT+ with built-in HAILO Accelerator (13 TOPS AI Power)
- 1 x PCIe FPC Cable (30mm), pre-assembled on the HAT+
- 4 x 2×20 16mm stacking header
- 4 x M2.5 Spacer and screw set
- 4 x M2.5 long screw
There are no reviews yet.