Machine vision has long been a mainstay of science fiction, so much so that many consumers may think that most of the industrial world is completely vision-enabled. That isn't quite true yet, but machine vision has actually played an important role in improving manufacturing operations since the 1980s.
As the technology progressed out of research laboratories and into practical implementations and as hardware and software underpinning have advanced, machine vision capabilities and applications have multiplied. The manufacturing marketplace continues to apply machine vision in new ways to improve performance and quality. In large part this is due to high-performance devices which can be economically applied to solve a variety of problems. Another major contributing factor to increased implementation is ease of use, as new solutions are much simpler to deploy and support than older vision hardware and software.
Some of the first machine vision applications involved rudimentary edge or spacing detection, but in the 1980s these evolved into systems that could read two-dimensional symbols and labels when properly presented to vision detection hardware. First generation "smart cameras" were limited and difficult to configure, but users were hungry for the flexibility this technology could offer over and above traditional sensors, so they put up with many initial implementation difficulties.
The latest generation of hardware and software can achieve high spreed three-dimensional analysis. In many ways this drives a virtuous circle of improvements and adoption as Forbes.com (Reference 1) observes "the rapid advancement of machine vision technology as the main reason companies seem to be buying more equipment. Companies can now buy cameras the size of quarters that can capture and process high-quality footage that just three years ago wouldn't have been possible." Better vision systems increase use in manufacturing applications, increased demand spurs vision system supplies to improve their products.
Another newsworthy and relevant application of machine vision is for self-driving cars and unmanned vehicle (drone) operations. These represent a much more cutting-edge application of vision than is used by manufacturing industries, but the underlying principles and needs are the same. Although manufacturing applications aren't widespread just yet, drones are being used now to perform inspections of assets such as electrical transmission lines (Reference 2) and can be deployed for inspection of other assets such as pipelines, tank farms, and water/wastewater distribution and collection systems.
This White Paper will address the latest technology trends in machine vision and show how these trends are providing benefits to manufacturers worldwide. high performance hardware options are making it easier than ever to install vision systems and different networking architectures have emerged to best serve a wide variety of diverse applications. Standards are emerging to ease integration, and software is becoming easier to set up and use.
While no single approach is ideal for all needs, a fundamental objective is to reduce the development and deployment effort by end users. Advanced hardware and software deliver on this goal, minimizing life cycle support and the total cost of ownership.