AI-at-the-edge computing

Edge AI computers run AI inference – object detection, classification, measurement or anomaly detection – right next to the cameras and sensors instead of in the cloud. That gives response times in milliseconds, less network traffic and data that stays on site. SCN supplies edge AI systems from AAEON, Axiomtek, Diamond Systems and IEI: fanless box PCs and industrial PCs with NVIDIA Jetson (Nano, Xavier NX, Orin Nano, Orin NX, Thor), Intel Core Ultra with a built-in NPU or a Hailo-8 accelerator, JetBox carrier boards and developer kits, an IP67 model and 1U/2U edge AI servers. They are used for machine vision, quality inspection, video analytics, autonomous mobile robots (AMR) and in-vehicle systems. Our engineers help you choose the right AI performance (TOPS), camera interfaces, I/O and enclosure.

Solution areas

Product categories

Filter by attributes

Acceleration

Accuracy (°)

Accuracy (%)

Approvals

Auxiliary contact

Built-in I/O

Capacity

Circuits

Coil Voltage

Communication Interface

Switching output

Contact config

Mechanical cycles

Continuous current

Control voltage DC

Control voltage AC

CPU performance

Data logging

DIN-rail

Electrical connection

Number of poles

Coding

Connector design

Cable jacket

Expansion slots

Fieldbus/Interface

Flange size

Function

Housing

I/O slave module

IP Classification

Load Cell Type

Material

Material in contact with medium

Max current DC

Max current AC

Max operating speed

Max system voltage DC

Max system voltage AC

Measuring range (mm)

Measuring range (bar)

Measuring range (kg)

Measuring range (°C)

Measuring range (°)

Torque

Power (kW)

Digital inputs

Digital outputs

Analog inputs

Analog outputs

Measuring range (l/min)

Measuring range (kHz)

Measuring technology

Measured variable

Memory

Modules

Mounting

Nominal diameter

Number of ports

Operating system

Outer Diameter

Output Voltage

Output current

Efficiency

Phase

Photocell type

Positions

Pressure type

Pressure limit PSI

Process connection

Probe length

RAID

Rated power

Remote Access

Resolution Multiturn

Resolution Singleturn

Screen size

Shaft Diameter

Storage

Stroke length

Supply voltage DC

Supply voltage AC

Temperature range

Form Factor

Touch

Type of Switch

Voltage

dB(A)

Brands

Edge AI computers for vision, robotics and vehicles

AI-at-the-edge computing means running AI inference – object detection, classification, measurement or anomaly detection – on a computer installed next to the cameras, sensors and machines that produce the data, instead of sending everything to the cloud. The result is response times in milliseconds, less network traffic and data that stays on site. SCN's range covers NVIDIA Jetson systems (Nano, Xavier NX, Orin Nano, Orin NX, AGX Orin and Thor) from AAEON, Axiomtek and Diamond Systems, fanless PCs with Intel Core Ultra and a built-in NPU, a palm-size system with a Hailo-8 accelerator and 1U/2U edge AI servers from IEI for heavier workloads.

Typical applications are machine vision and quality inspection, video analytics and traffic monitoring, autonomous mobile robots (AMR), in-vehicle systems and predictive maintenance. For general-purpose industrial computing without AI, see all box PCs.

How to choose: start with the AI performance you need (TOPS) and the number and type of cameras – GigE/PoE, GMSL2 or MIPI CSI. Then check the environment: fanless design, operating temperature, IP67 protection, wide-range DC input and CAN bus for vehicles. Finally decide between a ready-to-deploy box PC, a carrier board for your own enclosure or a developer kit for prototyping. SCN's engineers help you size the system and can also supply storage, cameras and accessories.

Frequently asked questions

What is edge AI?

Edge AI means running a trained AI model directly on a computer next to the camera, sensor or machine instead of in the cloud. The computer analyses images or signals on site and acts within milliseconds, even without a network connection. Typical uses are quality inspection, people and vehicle counting and robot guidance.

Which NVIDIA Jetson module do I need?

It depends on the number of camera streams and the size of the model. Jetson Orin Nano (up to 67 TOPS) handles one or a few streams, Orin NX (up to 157 TOPS) multi-camera vision and mobile robots, and AGX Orin (up to 275 TOPS) many parallel video streams and larger models. Jetson Thor is built for advanced robotics.

When is edge AI better than cloud AI?

When decisions must be fast, the connection is unreliable or the data must stay on site. A reject signal on a production line, a robot avoiding an obstacle or a camera in a vehicle cannot wait for a round trip to the cloud. The cloud remains the right place to train models and run analyses that are not time-critical.

Which edge AI computer works in vehicles and outdoors?

For vehicles, choose a model with wide-range DC input, CAN bus and GMSL2 camera inputs. For outdoor or wash-down installations, the AIE810-ONX is IP67-rated with M12 connectors. Developer kits from Axiomtek and carrier boards from Diamond Systems make it easy to test your model before choosing the production system.

Not sure which model you need?

Get help choosing
Read more in our guides

Facts and advice that make the choice easier.

Edge computing and AI – how to choose an industrial computerIP ratings explained – IP65, IP67, IP68 and IP69KIndustrial PC – choosing a box PC, panel PC or rack PCAll guides