EMPOWERING AI AT THE EDGE

From AI Acceleration Modules,
Inference Systems to Software

AIoT will serve as the next major growth engine across the business world. To help customers jumpstart their AI deployment and realize AI’s full potential, Advantech provides hardware and software integrated solutions to empower AI on the edge. Our solutions combine the hardware AI engine, inference computing platform, and AI featured software with industrial reliability. Advantech Edge AI solutions enable deep learning across different industries such as drones, AGV(automated guided vehicle), retail, robotic AOI, medical imaging, traffic monitoring and more.

Edge AI on_the_Edge01

Complete Solutions

Ranges from acceleration modules and inference systems to solution ready packages

Edge AI on_the_Edge02

GUI-based AI Toolkit

Edge AI Suite helps with features and power across various neural networks based on OpenVINO toolkit

Edge AI on_the_Edge03

IoT Management

With IoT device operation management built-in, our solutions help easily manage connected devices

Getting Started with Advantech AI Solutions

Edge AI started_3

Ready Edge AI Solutions

エッジAI加速モジュール

Edge AI
Acceleration Modules

Advantech provides low power add-on modules in different form factors including m.2, miniPCIe, and PCIex4 and can be easily integrated to your existing systems effortlessly while simultaneously accelerating computer vision and deep learning inference on the edge.

エッジAI推論システム

Edge AI
Inference Systems

Advantech provides powerful AIR series inference systems to enable deep learning inference in defect inspection, medical imaging, roadside monitoring applications and more.

Edge AI Suite Built-in

To help customers rapidly enable deep learning inferences on their edge devices with Advantech Edge AI products, Advantech offers Edge AI Suite. With Intel’s Distribution of OpenVINO™ toolkit and pre-trained models integrated, Edge AI Suite provides a deep learning model optimizer, an inference engine, pre-trained models, and a friendly GUI toolkit.

Edge AI Suite_01

OpenVINO Toolkit

  • OpenVINO R3.1 integrated
  • Boosted inference performance
  • Heterogeneous executions across CPU/VPU/GPU to optimize workloads
Edge AI Suite_02

Pre-trained Models

  • Rapid object detection
    (YOLOv3)
  • Facial recognition
  • Human pose estimation
Edge AI Suite_03

Deployment Wizard

  • Quick start tutorial
  • Supports various deep learning frameworks
  • CPU/VPU/Memory monitoring
Edge AI Suite_04

3rd Party AI SDK

  • Facial recognition
  • Vehicle classification
  • License plate recognition

Function Highlight

Quick Start

  • GUI-based operations
  • Real-time and stored video analytics
  • Supports Intel CPU/GPU/VPU accelerations

Inference Engine/Application

  • Matching and auto-labeling technologies integrated
  • Performance index
  • Source code provision

System Monitoring

  • Visualized dashboard
  • Benchmark evaluation
  • CPU/VPU/GPU and memory status

Model Launcher

  • Leverages OpenVINO model optimization
  • Supports various model frameworks
  • Model conversion parameter and sample codes offered

AI Use Cases

AI-Powered Smart Street Lights
Cities are illuminated by thousands of street lights that help drivers and pedestrians find their way home safely. Equipped with AI technology, street lights can be used to not only light up roads, but also support city services such as air quality, humidity, and temperature monitoring, as well as traffic and parking management.

Solution

The customer had already implemented Advantech’s edge computing platform to serve as a gateway computer for collecting data, including temperature and humidity levels, to facilitate operational analysis. Recently, the platform was upgraded with Advantech’s VEGA-320 edge AI acceleration modules embedded into the cameras for analyzing multiple streams of video data and handling real-time tasks such as locating available parking spaces. The customer estimated that this solution increases the utilization of roadside parking by at least 10% to 15%.

 

Key Benefit

• Compact, low-power design with minimal installation effort
• Video analysis at the edge for instant response
• Saves bandwidth resources and costs

AI-Based AGV Navigation
Automated guided vehicles (AGV) are used to transport and handle goods predictably and reliably. With the integration of AI-based navigation, the optimal route is determined in real time, effectively increasing the runtime efficiency of AGVs for faster and smarter operations.

Solution

For vision-guided AGVs equipped with cameras, Advantech’s VEGA-330 edge AI acceleration mini-PCIe module acts as a vision analytics engine, optimizing image processing and local inference for improved route planning and collision avoidance. This module is capable of processing and analyzing images captured by the camera to facilitate real-time AI-based navigation. By optimizing route planning, AI-based navigation increases the AGV runtime efficiency by up to 20%.

 

Key Benefit

• Compact design with standard interface for easy integration
• Low power consumption extends the battery life
• Enables AI-based vision-guided navigation

AI Facial Recognition for Retail
The accuracy of facial recognition algorithms has improved significantly in the past few years. AI facial recognition can help retailers proactively prevent shoplifting and enhance customer service.

Solution

The company required a hardware and software integrated solution to enhance its shoplifting prevention and customer service via AI facial recognition. The AIR-100 inference system integrated with FaceView can detect customers’ gender, age, and mood in real-time with 99.85% accuracy rating, enabling further analysis and precision marketing. The solution can also identify VIP customers or people with a shoplifting record, and send instant notifications to staff.

 

Key Benefit

• High recognition accuracy
• Plug & play functionality, no training required
• Provides functional APIs for easy integration

Robotic AOI Defect Inspection
Manufacturing quality controls have long relied on visual inspection. Traditional machine vision systems may fail to distinguish defects due to the variability and deviation between visually similar parts. By leveraging AI deep learning technology, this problem can be overcome and overall detection accuracy improved

Solution

Our customer is a robotic visual equipment builder. Their defect inspection solution combines robotic arms with AI capability to detect defects such as air bubble and crack among enamel coated products. Advantech AIR-300 AI system, powered by Intel 65W Core i7 quad core processor and a NVIDIA GPU card, enables real time inferencing and continuous training. It guarantees a high density image processing in real time and parallel computation to accelerating model training time. It also provide high bandwidth and storage capacity to collect captured images from multiple product lines and store massive training datasets.

 

Key Benefit

• Intel® Core™ i7 quad-core processor with NVidia 2080Ti GPU card
• Four GbE ports offer sufficient bandwidth
• Four 2.5” SATA III drive bays offer 20TB storage capacity

Download our brochure to
see our full range of solutions

Edge AI_download