Machine vision opens up new markets
Over the years, "machine vision" has become a well-established term describing the technology's main area of application in the manufacturing industry. However, the components and systems originally developed for the industrial sector are increasingly being used in non-industrial applications.
For many years non-industrial applications have been listed in the relevant market statistics as fields of application however, they have always lagged far behind the industrial manufacturing sectors that have historically been the most important for the machine vision market, such as the automotive industry, general engineering, electronics production, the pharmaceutical and chemical industries, food production along with many other industries.
For some time now, the proportion of turnover generated in non-industrial applications has grown much faster than in the industrial segments making it an increasingly attractive market sector for machine vision** companies.
There are several reasons for this development. On the one hand, the prices for machine vision technology have fallen considerably in recent years, as the manufacturers of components benefit from price developments in the consumer market, among other things the move to lower cost CMOS based image sensors and powerful embedded processors driven by the developments from mobile phone market. On the other hand, the rapid development of high-speed networks and deep learning enable easier deployment in the wider environment, whereas in the past the technology was limited to small areas of a constrained factory. This has enabled the players to take advantage of the new technical features of the technology to expand into these new markets and improve their growth potential, as shown by the following examples.
Automated track monitoring
In terms of robustness, machine vision systems used in traffic engineering face similar conditions needed in industrial environments. This is particularly true for railway applications, where vibrations, wide temperature ranges and changing weather conditions often result in extremely harsh environmental conditions.
With an automated track monitoring system, London Underground Limited (LUL), part of Transport for London, has been monitoring the wheel-rail interface and the tracks of the London Underground railway network for many years looking for possible damage.
For this purpose, cameras in specialist enclosures capable of operating in the harsh environmental conditions were mounted on the bogey and at the end of the carriage in order to record and evaluate image data during operation. By acquiring this data during normal timetable periods, more time is available during the 4-hour night closure period to maintain the network’s 1000 km of track, minimising any disruption to services and helping to make the new weekend extended operating times possible.
The EMC-resistant enclosures certified to IP65 are shock rated to 5G for continuous load (shaking) and 50G for drop. In this challenging project, high-speed cameras, IR filters and high-speed LED IR pulsed illumination, as well as specially developed data recording and processing methods, form the basis for a more efficient monitoring of the London rail network.
Sports and entertainment are further areas where vision systems are increasingly being used. In professional matches in football, basketball, cricket and many other sports, both players and equipment are tracked to allow calculation of players’ total running performance including average and maximum speeds, number and intensity of sprints and the distance covered. Based on this data, football coaches can decide when to replace which player, thereby reducing the risk of injury to tired athletes.
Such systems also use this type of image data to clarify whether a ball still touches the goal line or has completely crossed it, and are therefore ideally suited for verifying referee decisions in sports such as football, tennis and cricket. In another application highly sophisticated 3D vision systems identify the precise landing position of darts on the dartboard to within a fraction of a millimetre and immediately display the score.
In many cases, sports equipment is travelling at very high speeds. Cricket balls can be bowled at speeds approaching 100 mph, while the world record for tennis serves even exceeds 160 mph. The challenges for imaging technology in these applications are extremely high in terms of speed and resolution. In addition, as most of the sports are conducted outside, the systems must be able to accommodate varying illumination conditions.
One of the most impressive examples of high-speed ball tracking comes in the form of “RoboKeeper”, an automated goalkeeper which is marketed as a visitor attraction at major events. Penalty shooters are invited to compete against the RoboKeeper and try to score a goal. In most cases this is almost impossible, because a powerful combination of image processing and actuators ensures that the ball's trajectory is detected within a few milliseconds and the automated goalkeeper prevents the ball from crossing the line. To compensate for variations in lighting conditions for outside events, the camera systems are equipped with lenses with an automatic iris controller and a motorised iris.
Hunting world records with vision
The British Bloodhound project is currently on a world record hunt with the support of machine vision: The jet and rocket powered car has made significant progress towards its goal of increasing the speed from around 760 mph to 1,000 mph (around 1,600 km/h) and achieving a new world land speed record. Up to 25 cameras can be located at strategic points on the vehicle. During the test drives and record attempts, their data will be used to ensure the fault-free functioning of components, some of which are safety-relevant, such as the correct ignition of the rocket propulsion system, the tire-to-ground contact or the brake parachute.
Vision in the arts
A more peaceful environment is the setting for a system that allows art lovers to discover completely new and unexpected insights into the world of art. Visual artworks can be viewed in a completely new way and with unprecedented levels of detail. The technology creates interactive 5D images and movies that allow a special and emotional experience when viewing an artwork on screen. Depending on the physical size of the artwork, up to a hundred thousand images are taken by a special scanner using different light sources and spectra, including ultraviolet and infrared illumination. The result is then formatted for online visualisation.
Art lovers can then visualise the digital twin of an artwork online. In particular, the interactive representation of the 5D images in real time allows a new form of enjoyment from art. The technology also simplifies the creation of interactive, audiovisual guides or electronic catalogues and can also be used to ensure the security of the valuable treasures: Thanks to the high level of precision, the system creates an individual fingerprint of the artwork, which makes forgery practically impossible or can be used to prove possible damage in insurance cases.
Wind turbines stop for birds
Machine vision is even being used to protect wildlife. With the increasingly strict wildlife conservation regulations, wind turbines have to be switched off when a protected species of bird approaches them. With the help of artificial intelligence and suitable imaging components, a recently introduced machine vision system banishes the danger of birds being struck by wind turbine blades and minimises expensive downtimes.
The system recognises and localises large birds of prey and enables their flight path to be tracked in order to shut down wind turbines only when protected birds move within a certain distance of the turbines. The recognition of birds and their classification for the subsequent decision as to whether the bird is a protected species is an extremely demanding task which has been solved under difficult conditions due to outdoor operation.
To capture the entire surrounding area with a 360° all-round view, six industrial colour cameras with resolutions of up to 20 megapixels with special wide-angle lenses in weatherproof protective enclosures are used. However, the evaluation of the captured images and the reliable recognition of endangered birds represent the actual skill in this application Without powerful deep learning methods to train several hundred thousand images with positive and negative instances and the relevant know-how in handling these innovative methods, it would not have been possible to solve this challenging task.
Vision components and systems from STEMMER IMAGING play a decisive role in all the examples mentioned. The solutions described have been implemented in close cooperation with the company's partners and customers and are entirely based on products designed for use in industrial applications. However, due to their price and performance, they are also ideally suited for non-industrial applications. The company is a well-established leader in the field of machine vision and aims to significantly expand its activities in the evolving non-industrial applications.
"The example of the wind turbines is an impressive illustration of the possibilities that Artificial Vision has to offer in a wide range of applications."