0021 MPSAI EN

Keywords

Raspberry Pi 5, PCIe, Driver installation, 2230, 2242, 2280 NVMe SSD, SSD X1, AI, HAILO 8L, Camera, OV5647, Image Recognition

I. Introduction

MPSAI is a dual-interface expansion board that supports SSD and AI modules from HAILO. MPSAI employs a PCIe Switch chip to bifurcate the PCIe signal, with one path linking to an SSD and the other to an AI module. SSD can be used for booting, as well as for mass storage of training videos or processed videos.

II. Hardware Spec

1. Designed for the Raspberry Pi 5, it uses a 16-pin FPC to connect to the Raspberry Pi 5's PCIe interface.

2. Two PCIe interfaces via a PCIe Switch.

3. One PCIe M.2 E key slot, 2242 size, used for connecting the AI module HAilo8，with the official Raspberry Pi OS loading the driver.

4. One PCIe M.2 M key slot, for connecting NVMe SSD (PCIe X1, Gen2 mode only). Supports 2230/2242/2280 size SSDs (defaulting to welded mounting posts for 2280 size). It does not support SSDs that use NGFF and SATA protocols.

5. The SSD can be used for data storage and system booting. When booting from the SSD, the Raspberry Pi official firmware needs to be version 2024.5.13 or later.

6. The 40-pin GPIO section adopts an unobstructed design on the top. It can be connected via DuPont cable, without affecting the use of the Raspberry Pi 5's own GPIO pins.

7. Size：86*56mm.Gold immersion PCB process, lead-free production, certified by UL, compliant with ROHS standards, and has a fire rating of 94V-0.

8. It uses four M2.5 mounting holes that align with the Raspberry Pi 5 mounting holes.

9. It is compatible with an optional aluminum alloy enclosure.

III. AI (HAILO 8L) work with Raspberry Pi OS

The version of the Raspberry Pi OS is: 2024-07-04-raspios-bookworm-arm64.img.xz

You can download the Raspberry Pi OS in:

https://www.raspberrypi.com/software/operating-systems/#raspberry-pi-os-64-bit

The OS is flashed onto the TF card or SSD.

For the flashing method, please refer to:

How to flash OS

3.1 Overview

In various practical applications, the demand for camera image detection and recognition is gradually increasing, and there are many AI-based solutions available on the market. The solution demonstrated in this article involves the OV5647 camera, the MPSAI, and the HAILO 8L AI module paired with the Raspberry Pi 5.

Note: The operation requires a stable internet connection to the regions out of China (you may need to find your own method). Otherwise, many files may not download or may only partially download, which could ultimately lead to the code failing to run.

3.2 Update system

Note: The following operational steps apply whether the system is booted from a TF card or an SSD.

3.2.1 Boot the Raspberry Pi OS and run commands in the terminal:

sudo apt update

sudo apt full-upgrade

3.2.2 Run sudo rpi-eeprom-updateto check the Raspberry Pi firmware and ensure that the firmware date is after December 6, 2023:

As shown in the above figure, the firmware date is June 5, 2024.

3.2.3 Run sudo raspi-config ,select the corresponding options in sequence according to the following diagram:

After making your selection and exiting, the system will automatically restart. If it does not restart automatically, please manually execute sudo reboot to restart the system.

3.3 Install the depends required for the AI module

We are using the OV5647 camera, connected to the CSI interface. After the OS starts, open the terminal and execute the following commands:

sudo nano /boot/firmware/config.txt

Insert the following lines at the end of the file:

dtoverlay=ov5647,cam0

Or:

dtoverlay=ov5647,cam1

Please determine based on the actual connected CSI interface.

Run sudo apt install hailo-all, install the depends required for the AI module:

After completing this, execute sudo reboot to restart the system and apply these settings.

Run hailortcli fw-control identify, If the following output appears, it indicates that the depends for HAILO 8L have been successfully installed:

Finally, execute rpicam-hello -t 10s to turn on the camera and bring up the preview window, ensuring that the camera is functioning normally.

3.4 Image recognition and detection

Run:

git clone --depth 1 https://github.com/raspberrypi/rpicam-apps.git ~/rpicam-apps

Clone the library of rpicam-apps:

After cloning is completed, you can begin image recognition and detection:

rpicam-hello -t 0 --post-process-file ~/rpicam-apps/assets/hailo_yolov6_inference.json --lores-width 640 --lores-height 640

You should downloading a large amount of data from the internet out of China to use the testing of this development board, so users can contact our company after buying this development board to obtain a pre-configured Raspberry Pi OS image.

IV. Using SSD in the Raspberry Pi OS

An SSD can be used as OS storage. For the method of flashing the system to the SSD, please refer to: Click here.

For basic operations and tests on the SSD, we can refer to the following link:

Use the SSD for storage expansion (Raspberry Pi OS)

Perform partitioning and other operations on the SSD

Adjust the peripheral boot order

Install the SSD speed testing software hdparm

V. Using SSD in the Ubuntu

The version of the Ubuntu is: ubuntu-24.04-preinstalled-desktop-arm64+raspi.img.xz

You can download the Ubuntu in:

https://ubuntu.com/download/raspberry-pi

For basic operations and tests on the SSD, we can refer to the following link, apart from the SSD being used for storage expansion, which operates slightly differently under Raspberry Pi OS, the rest is basically the same.

Use the SSD for storage expansion(Ubuntu system)

Perform partitioning and other operations on the SSD

Adjust the peripheral boot order

Install the SSD speed testing software hdparm