3-D multiscanning, ease of use and connectivity, and lower costs are driving the growth of smart cameras and vision sensors in factory automation.
Anders Murhed and Mats Gökstorp, Sick IVP
In factory automation, the constant need for productivity and quality improvement is generating a strong demand for sensor solutions based on smart cameras and vision sensors. In choosing a vision system, the foremost consideration is ease of integration into the production line.
New systems are adapting more and more to industrial standards for programmable logic controller communications field buses. Open connectivity software standards are helping to integrate the camera system into the supervising systems and operator panels. Also important is ensuring that a vision system is robust enough for its environment.
New vision systems are bringing three-dimensional capability to factory automation, a welcome development because 2-D technology does not always produce good results.
Today, 3-D machine vision systems can determine real shape in inspection applications, adding height and shape data that can be critical when classifying and evaluating an object (Figure 1). Improved hardware performance from advances such as those in high-speed CMOS sensors and field-programmable gate array technology has made 3-D solutions possible.
Figure 1. A 3-D vision system shows the height and shape of an object, such as the brake pad shown here. The height of the brake pad is color-coded. Notice that the dent on the surface is easy to detect.
The 3-D image is generated by projecting a laser line across the object. By observing the object from an angle with respect to the laser line and finding the position of the laser line on the sensor, simple geometrical triangulation can compute the three dimensions. High-speed systems now can provide thousands of 3-D profiles per second, generating images at almost the same rate as 2-D processes.
It also is now possible to perform robot-based picking of varying and fragile items (Figure 2). Three-dimensional object measurement means that the item can be picked precisely, independent of height, tilt or rotational variations. In addition, 3-D vision measurements do notrequire as much contrast between the object and background, and color variations of the objects are not critical. For example, it is easy to locate a black object on a black conveyor belt, which would not be possible with a 2-D system.
Figure 2. This 3-D camera is used in a system that handles fragile products that vary in size. With a 2-D camera, a picker can miss an item or even destroy it. But with coordinate transformation and conveyor tracking using 3-D, the system evaluates the volume and shape of the bread and gives the robot the position of any misshapen loaves. The robot then moves a rejected loaf off the conveyor.
The Multiscan measurement mode for 3-D cameras, created by Sick IVP of Linköping, Sweden, can complement 3-D systems. It enables one camera to simultaneously acquire a number of important properties, such as 3-D, gray scale and scatter (Figure 3). The images are captured when the object passes the camera in line-scan mode, and they are transferred to a PC via Camera Link or Gigabit Ethernet for easy system integration. Previously, in certain applications, multiple cameras were required to produce the same result.
Figure 3. The Ranger camera uses a Multiscan mode to evaluate blister packs of pills in this pharmaceutical application. The camera captures several properties, such as 3-D, scatter and gray scale, in the same scan at up to 10,000 Multiscan blocks per second.
Some traditional sensor companies are jumping into the market rather than working with vision companies or integrators, and this trend will trigger a change in the industry.
As the technology becomes easier to use, small vision sensor companies will have difficulty keeping up with the development and production power of their larger counterparts. In addition, globalization of the industry requires a worldwide sales organization to meet customer needs. At the same time, the small companies will provide local solutions and continue to push the limits of technology, creating a healthy co-existence that will continue to drive the industry.
Cost reductions in hardware enable powerful machine vision solutions at higher integration and lower cost levels than before. The emergence of smart cameras is one example. The trend continues by providing vision sensors in the segment between smart cameras and industrial sensors.
The most important issue is packaging the advanced machine vision technology into easy-to-use products. Sick IVP is doing this mainly through software, and the challenge is to provide complex machine vision function in an easy-to-use package for a large market. Industrial sensor companies are well-positioned for this venture with their experience in offering both straightforward light switches and advanced vision technology.
The vision sensors and smart cameras make it easier for machine builders to include vision in their production devices and for end users to add inspection directly on existing production lines. Not surprisingly, good growth is expected for application-specific vision sensors. The advantage with a sensor is that the function is well-defined, and little engineering effort is needed to get it up and running. Shorter sales processes and lower costs will make these products successful for automation integrators with little vision experience, as well as for end users. Some examples of vision sensors include code reading, optical character recognition/optical character verification, color sorting and contour matching.
Visualizing the future
GigE Vision and Camera Link will be the interfaces for high-end PC systems in industrial applications. Solutions based on mid- and low-range PC systems using FireWire and USB will decrease and move to the smart camera segment. GenICam is a software standard that makes it easy to connect cameras to PC-based systems. GenICam and GigE will make it easy to exchange cameras in a solution, undoubtedly increasing competition for those companies selling cameras for PC systems.
Because 3-D vision in general is growing, there will be more 3-D offerings in smart cameras, which are becoming more reliable, powerful and easy to use. Therefore, some of the midrange, PC-based vision solutions can use smart cameras for lower cost and reduced complexity. The main growth, however, will come in areas where vision was not possible before because of complexity and cost. More machine builders will integrate smart cameras as standard components on production units, offering their customers higher yield and quality.
PC-based systems also are growing, but most likely not as fast as the vision sensors and smart cameras. Complex inspection tasks still will be best solved by PC-based systems, as when several cameras are needed for the inspection task. The processing power also is higher in a PC than in embedded smart camera processors. These can include automated online inspection systems in electronic production lines, wood quality inspection and other applications where several cameras often are used to measure various features.
Meet the authors
Mats Gökstorp is general manager of Sick IVP in Linköping, Sweden; e-mail: email@example.com.
Anders Murhed is marketing and sales manager of Sick IVP; e-mail: firstname.lastname@example.org.