- Google Edge TPU ML accelerator
- NXP i.MX 8M SoC (Quad-Core)
- 1 GB LPDDR4x
- 1x GbE PoE/PD LAN port
- Operating temp.: -5 °C ~ 50 °C
Description
BOXER-8521AI – AI@Edge Embedded Platform
Winner of the 2021 Taiwan Excellence Award, the BOXER-8521AI combines installation flexibility with advanced connectivity features. It includes a PoE-PD port, allowing the system to be placed far from the power source while maintaining internet connectivity and remote monitoring through a single cable. Additionally, the system can operate with both the PoE-PD port and DC input, ensuring uninterrupted functionality even if one power source fails.
AI Edge Computing with Google Edge TPU
The BOXER-8521AI is designed for AI edge computing, powered by the Google Edge TPU system-on-module (SoM). This module integrates the NXP i.MX 8M SoC (Quad Cortex-A53, Cortex-M4F) with the Google Edge TPU coprocessor, delivering up to 4 TOPS at a highly efficient 2 TOPS per watt. The system’s robust chassis and versatile I/O workload make it ideal for demanding edge computing tasks.
Rapid Development of AI Models
With the Google Edge TPU, developers can utilize Google’s AutoML Vision Edge software suite to quickly create and train AI models. This innovative tool leverages Google’s cloud services for training, then exports the trained model to the BOXER-8521AI for on-device processing. TensorFlow Lite support ensures efficient AI model execution, improving system performance and processing speeds.
Rugged and Reliable Design
Built with a durable, fanless design, the BOXER-8521AI protects against dust and contaminants, ensuring long-term reliability. It operates within a temperature range of -5°C to 50°C without performance loss. The system includes flexible I/O options such as COM, HDMI, two USB3.2 Gen 1, and two USB2.0 ports. Additionally, a 40-pin multi-I/O port allows integration with sensors and controllers, including cameras, temperature sensors, and other devices.
Specifications
| BOXER-8521AI | |
|---|---|
| SYSTEM | |
| CPU | NXP i.MX 8M SoC (Quad-Core Cortex-A53, plus Cortex-M4F) AI Accelerator: Google Edge TPU ML Accelerator Coprocessor |
| RAM | 1 GB LPDDR4x |
| Storage | 8GB eMMC |
| Operating System | OS Support: ACLinux 4.0 (compatible with Debian 10) |
| INTERFACE | |
| Ethernet | 2x RJ-45 for PoE/PD (802.3at) |
| USB | 2x USB Type A for 3.2 Gen 1 2x USB Type A for 2.0 1x OS Flash Port (Micro-USB) |
| COM | 1x DB-9 for RS-232/485 |
| Digital I/O | 1x 40-Pin I/O (GPIO/I2C/SAI/SPI/UART/PWM) |
| Video | 1x HDMI 2.0a |
| Audio | 1x Mic-In 1x Line-Out |
| Power | 1x 12V DC (lockable DC connector) |
| Other | 1x MicroSD Slot |
| ENVIRONMENTAL | |
| Power Supply | PoE/PD (802.3at) 1x 12V DC (lockable DC connector) |
| Remote Control & LED | 1x Power LED |
| Operating Temperature | -5°C ~ 50°C |
| Storage Temperature | -40°C ~ 85°C |
| Vibration/Shock Resistance | Vibration: Random, 3.5Grm / 5 ~ 500Hz Shock: 50G peak acceleration (11 msec duration, eMMC, microSD) |
| Humidity | 5 ~ 95% @ 40°C, non-condensing |
| Dimensions | 175.8 x 100 x 39 mm |
| Weight | 1.05 kg (gross) 0.58 kg (net) |
| Mounting | Wall-Mount Kit |
| Certifications | CE/FCC Class A |
