Industrial IoT & Machine Vision Solution based on AM5718

Industrial IoT and Machine Vision

1. Definition

Machine vision is a kindly of comprehensive technology, including image processing, mechanical engineering technology, controlling, optical imaging, sensor, analogy and digital video processing, computer hardware and software engineering (image enhancement and analysis algorithm, image card, I/ O card, etc.)

A classic machine vision system consists of image capturing, light source system, image digitalizing, digital image processing module, intelligent decision supporting module and mechanical control execution module.

The basic feature of machine vision system is that it can improve production flexibility and automation. In some industrial working circumstance where is dangerous for manual operating or occasion that manual vision is not enough for requirement, in these cases, it will be much more convenient to take machine vision to replace manual vision. Meanwhile, in process of repeated mass production, machine vision system inspection can improve production efficiency and automation to a large extent.

Operational Principle

Machine vision inspection system will convert the detected items to image signal by CDD or COMS cameras, and pass the signal to special image processing system, then according to pixel distribution, brightness and color to convert it to digital signal, and image processing system will compute the signals to extract features of the target, such as area, quantity, position, length, and then output result according to predetermined terms and other related terms such as dimensions, angle, quantity, pass or not, with or without.

2. Machine Vision Features

a. High accuracy. It can accurately measure target objects according to selected scenes and resolution of sensors.

b. High continity. Contactless mearsurement can help protect product's lifetime and reduce workload of the machine.

c. High flexibility. Machine vision can do various measurements.

TI AM5718 Applied in Machine Vision

AM5718 consists of a Cortex-A15 core with requency up to 1.5GHz, a DSP with requency up to 750MHz, two dual-core PRU with requency up to 200MHz and two Cortex-M4 cores with frequency up to 213MHz. It can be widely used in industrial communication, HMI, automation and control, high performance applications, and other general use.

Forlinx AM5718 Development Platforms

1. FET5718-C system on module: Designed based on TI AM5718 processors, and supported with Linux4.9.41、QT5.6

2. OK5718-C single board computer: Designed based on FET5718-C which is very convenient for testing purpose.

Machine Vision Application Based on FET5718-C


1. Integrated DSP+OPENCV environment

TI C66x VLIW floating-point DSP core. The ARM processor lets developers keep control functions separate from vision algorithms programmed on the DSP and coprocessors, thus reducing the complexity of the system software. Forlinx kindly provides Sobel, facial recognition DEMO, OpenCV+OpenCL+OpenGL DEMO.

2. PRU

  • a. CPU subsystem PRU-ICSS integrated with real-time communication protocols but no need of external ASIC or FPGA. PRU-ICSS can support Ethernet, field bus, interface and VRRP protocol applicable EtherCAT, Profibus and Profinet.
  • b. Real-time IO controlling, latency can as low as ns
  • c. Applicable for CNC, relay protection and power controlling.
  • d. Separate interrupt and internal RAM which allow it to visit peripherals.

3. Multiple camera interfaces

FET5718-C can support two 8-bit parallel camera interface, two 8/ 16/ 24-bit serial interface, available for DVP OV5640, MIPI OV5640 and USB camera for photo taking, previewing and recording. DVP OV5640 can auto focusing. And can support resolutions of 1920*1080, 1280*720, 640*480.

4. Compact size, powerful performance and low consumption

FET5718-C dimensions are 50*70mm which make it more compact than X86 machine, it can keep powerful performance even at low power consumption. Industrial grade temp width makes it can work well even in tough environment. Optimized files system provided by NXP simplized from 3GB to 1.6GB which shortens system booting time. Besides, SD card flashing solution can help users to improve mass production efficiency in system flashing but no need manual intervention.

Target Application

Industrial communication, HMI, automation& control, smart building, machine vision, medical image, measurement, car entertainment.