In this article, Sealevel Systems discusses the difference between machine vision and embedded vision systems, and the benefits embedded vision systems can offer in industries and markets that traditional machine vision systems cannot.
Embedded vision systems combine all the hardware and software components of machine vision – image capture, processing, and interpretation – into one device. By eliminating the need for an outside connection, embedded vision systems can be utilized in industries and markets that traditional machine vision systems cannot. Further, embedded vision systems have a significantly smaller footprint, which offers opportunities for applications that have stringent SWAP-C2 requirements.
Essentially, machine vision includes all applications in which machines can automatically capture, process, and interpret visual information from the world around them to make decisions. The primary use cases of machine vision are image-based inspection, process control, and robot guidance, primarily in industrial and manufacturing settings. For example, machine vision systems are used on assembly lines to identify defects and anomalies and remove these units from production.
In traditional machine vision systems, the image-capturing component – generally a camera or a sensor like a LIDAR – is connected to a server-class computer. In this system, the server-class computer has high-level processing capabilities and software that is programmed to handle image processing, interpretation, and output. Generally, in manufacturing settings, the output is a pass/fail decision, with defective or anomalous items being removed.
Embedded Vision Systems
Embedded vision systems perform many of the same tasks as machine vision but combine all the hardware and software components – image capture, processing, interpretation, and output – into one edge computing device. These systems are compact, and usually consist of a camera mounted directly onto an image processor.
Currently, embedded vision systems are more limited in their image processing capabilities compared to larger and more complex machine vision systems. However, as smaller and smaller processors and cameras are developed, the gap between machine vision and embedded vision will shrink. Further, as the cameras and image processors are currently tailored to each specific application, embedded vision systems tend to have a larger upfront cost than machine vision systems but have significantly lower energy requirements and lower operational costs.
While embedded vision technology is relatively new, it is already making an impact in several industries:
Automotive – Embedded vision technology has been incorporated into many Advanced Driver Assistance Systems (ADAS). This includes things like automatic emergency braking, collision warning, lane departure warning, blind spot warning, adaptive cruise control, and more. As both the European Union and the US requires ADAS technology on all new cars, it stands to reason that the use of embedded vision systems will increase.
Industrial Robotics – Embedded vision systems are being incorporated into robotics to improve performance and increase functionality in industrial robots. This includes robotics involved in a huge variety of applications: material handling, pick and place, inspection, quality control, warehouse management, and more.
Chemical and Pharmaceutical Manufacturing – Embedded vision systems in chemical and pharmaceutical manufacturing are currently being used in inspection tasks for quality assurance, especially at high speeds and volumes. As both chemical and pharmaceutical products are highly regulated, the use of vision-enabled robotics reduces production costs.
Medical – Devices with embedded vision technology are being used to improve the diagnostic capabilities of medical professionals. Vision systems have proven to be quite adept at identifying early warning signs for skin cancer.
The Future of Embedded Vision Systems
The main benefit of implementing embedded vision systems is their flexibility. As these systems are physically smaller, they offer significant SWAP-C2 benefits. As embedded vision systems are uncoupled from an external computer they open completely new imaging capabilities with endless commercial opportunities.