ARMxy BL370 | RK3562 Industrial Computer with 1TOPS NPU and 58,302 I/O Combos
No.BL370
ARMxy BL370 industrial computer powered by RK3562. 1TOPS NPU for edge AI, 4K video codec, and Linux-RT (<50µs jitter). Choose from 58,302 factory-built I/O combos. Ideal for machine vision, AGVs, and industrial automation. -40°C to +85°C wide temp.
BL370 is designed for applications that demand both real-time control and intelligent processing — from machine vision to predictive maintenance.
1TOPS NPU · Edge AI
Built-in neural processing unit (NPU) supports TensorFlow Lite, PyTorch Mobile, Caffe, MXNet. Run YOLOv5/8, OpenCV, and custom models at the edge — no cloud dependency.
4K Video Codec
H.265 4K@30fps decoding + H.264 1080P@60fps encoding. Perfect for machine vision, surveillance, and video analytics applications.
Real-Time Control
Linux-RT (PREEMPT_RT) kernel with <50µs jitter. Supports IGH EtherCAT Master (open-source) and CODESYS Runtime (licensed) for motion control.
RK3562 Quad-Core A53
22nm process, up to 2.0GHz. 4x Cortex-A53 + 1x Cortex-M0 (200MHz) for low-power co-processing. 8/16/32GB eMMC, 1/2/4GB LPDDR4X.
Modular Architecture
58,302 Factory-Built I/O Combinations
5 SOM boards × 11 X boards × 30+ Y boards × 3 wireless options = 58,302 possible configurations. Order your exact combination — we assemble and test before shipping.
5 SOM Boards11 X Boards30+ Y Boards58,302 Combinations
RS485/232/CAN
DI/DO/Relay
AI: 0-20mA/±10V
AO: 0-20mA/0-10V
PT100/PT1000
Thermocouple (J/K/T/E/R/S/B/N)
PWM/Pulse Counter
Example 1: High-Performance AI + Serial Aggregation (Y63 Exclusive)
Ideal for: Analog sensor acquisition + switch status monitoring with NPU edge intelligence
Selection Note: When Y63 (4x RS485/232) is selected, this Y-board occupies the slot exclusively — no second Y-board can be selected. Plan your serial port requirements accordingly.
All I/O boards are carefully selected and assembled before delivery — plug and use.
Edge AI Acceleration
1TOPS NPU for On-Device Intelligence
BL370 integrates a dedicated neural processing unit (NPU) that delivers 1 TOPS (trillion operations per second) of AI compute power. Run inference locally — no GPU, no cloud latency, no data privacy concerns.
✅ Supports INT4/INT8/INT16/FP16 quantization
✅ Compatible with TensorFlow, PyTorch, Caffe, MXNet, ONNX