NPC-7312 Edge AI Controller
A high-performance edge-AI controller for multi-camera vision, robotics and autonomous-machine workloads. The platform combines Jetson AGX Orin processing with six Ethernet ports, industrial serial/CAN connectivity and wireless expansion.
Deployment Capabilities
Local AI processing
Jetson AGX Orin resources can run supported detection, classification and tracking workloads near the data source.
Multi-camera aggregation
Six Ethernet interfaces provide a structured way to connect network cameras and distributed sensors.
Control-system integration
Serial, CAN and terminal I/O allow application results and equipment data to be exchanged with industrial systems.
Connected remote deployment
Wide-voltage input and wireless expansion suit stationary and mobile edge systems within the documented operating limits.
Technical Features
High-performance inference
Jetson AGX Orin provides GPU and tensor acceleration for multi-stream AI workloads. This can support local detection, classification and tracking without sending every video stream to a remote service.
Camera aggregation
Six Gigabit Ethernet ports support multiple network cameras and distributed sensors. Dedicated ports simplify camera separation, bandwidth planning and fault isolation in vision systems.
Industrial communication
RS-232/422/485, CAN and isolated terminal I/O connect machinery and vehicle interfaces. The interfaces allow inference results to trigger actions and exchange data with existing control systems.
Wireless expansion
M.2 options support Wi-Fi/Bluetooth and 4G/5G connectivity. Remote systems can combine local inference with central monitoring and fleet management.
Protected power input
9–36 VDC input with over-current and over-voltage protection suits mobile and industrial systems. The wide input range is useful in cabinets, mobile equipment and battery-powered machines.
Software ecosystem
Ubuntu and NVIDIA JetPack provide drivers, CUDA, AI libraries and deployment tools. The software stack provides tools for model integration, optimisation and deployment on the selected platform.
Specifications
| Parameter | Unit | Value / Description |
|---|---|---|
| COMPUTE | ||
| Compute module | — | NVIDIA Jetson AGX Orin. |
| CPU | — | Up to 12-core Arm Cortex-A78AE at up to 2.2 GHz, depending on module. |
| GPU / AI | — | Ampere GPU with up to 275 TOPS INT8 performance. |
| Memory | GB | 32 or 64. |
| Software | — | Ubuntu / NVIDIA JetPack SDK ecosystem. |
| I/O & EXPANSION | ||
| Ethernet | — | 6 × Gigabit Ethernet. |
| USB | — | 4 × USB 3.0 Type-A. |
| Serial | — | Industrial RS-232/RS-485 combinations. |
| CAN / digital I/O | — | CAN bus and digital I/O terminal interfaces. |
| Storage | — | M.2 NVMe SSD support. |
| Wireless | — | M.2 Wi-Fi/Bluetooth and 4G/5G options. |
| Security / deployment role | — | TPM 2.0 and GPS options. |
| MECHANICAL & POWER | ||
| Power input | VDC | 9–36 with over-current and over-voltage protection. |
| Dimensions | mm | Approximately 230 × 165 × 91.5. |
| Cooling | — | Fan-equipped industrial enclosure. |
| Application areas | — | Machine vision, robotics, sensor fusion and edge inference. |
Applications
Machine vision
Process multiple camera streams for inspection, defect detection, counting and classification close to the production line. Six Ethernet ports simplify camera and network segmentation.
Autonomous machines
Run perception, localisation and decision workloads for robots, mobile platforms and intelligent machinery. CAN, serial and digital I/O connect the AI system to machine controls.
Smart infrastructure
Analyse video and distributed sensor data locally for traffic, safety or facility applications. Edge inference reduces bandwidth and response latency.
ADAS development
Support camera, network and AI workloads for prototype vehicle and validation systems. Wide-voltage input and automotive-style interfaces suit mobile test environments.
Industrial analytics
Combine local inference with equipment data from Ethernet, serial and CAN networks. Results can be used for quality, anomaly detection and predictive-maintenance applications.
Research systems
Use NVIDIA JetPack, CUDA and AI libraries for algorithm development and deployment. The industrial enclosure provides a path from laboratory prototype to field evaluation.