Machine builders in this day and age are part of a fast-growing and rapidly changing industry sector. With vast improvements in commercial off-the-shelf products for measurement and control, as well as falling prices, machine builders can now take advantage of an assortment of powerful tools when implementing their designs, rather than rely on the age-old method of making a custom design for every application.

Machine vision can be used to verify placement of microchips on a PCB, or to verify the seal on a bag of potato chips.

Engineers looking to add machine vision to their industrial applications should find smart cameras a powerful addition to their toolboxes. Smart cameras combine high performance and powerful real-time processing, along with the I/O needed to easily integrate them into a new or existing design.

A smart camera combines an image sensor with a built-in processor, which allows inspections to run directly on the camera. Instead of returning images like a traditional camera, a smart camera returns results of inspections. This type of system, which can help lower overall system complexity, is typically easy to configure or program.

Machine designers are looking for a way to cut cost, complexity, and development time in their new designs. Smart cameras can fulfill these demands. Smart cameras eliminate the need for an external lighting controller that is typically seen in machine vision applications because the controller is built directly into the smart camera. This allows direct connectivity with current-driven light heads to help smart cameras source up to 500 mA of continuous DC current and up to 1 A strobed current. In addition to reducing the wiring needed, this cuts the cost of an additional piece of hardware. It also decreases development time because the application programming interface (API) for the lighting controller is built into the image acquisition API for the smart camera.

Machine vision software has become a powerful, easy-to-use piece of a machine builder’s toolbox.

Smart cameras are designed to perform many advanced functions as well as the basics that anyone in the machine vision field has grown to know and love (or lament, depending on the experiences with them). These include edge detection, geometric pattern matching, optical character recognition, and 2D barcode reading. This means that instead of integrating a sensor for a 2D code reading, and another sensor for object detection, and yet another sensor for optical character recognition, engineers can do everything on one smart camera, lowering the cost and complexity of the finished product.

Building a Machine Vision System

While some machine builders may be completely comfortable with programming their machine vision applications, others simply don’t have the time or desire to learn a programming language just for the machine vision aspect of their applications. With vision hardware and software, there are options for setting up an inspection.

The first option is to configure the inspection using software such as NI Vision Builder for Automated Inspection (AI), a no-programming, menu-driven machine vision package. This application gives those with little or no machine vision experience the ability to configure an inspection in a number of days, not weeks. This software is shipped with each NI Smart Camera.

The second option is to program the application using NI LabVIEW Real-Time and the NI Vision Development Module. This option is available for engineers who are familiar with the Lab-VIEW graphical programming language, or those who need more flexibility in a configurable environment.

Solution Applications

Smart cameras can be used in many manufacturing and packaging applications.

There are numerous applications where such a solution is optimal. A general application for machine builders is robot control. A common solution for pick-and-place applications is to have a proximity sensor fire when the desired object is in range and, as a result, the robot arm attempts to pick it up. This works well for a few applications, but some require more data than “Yes, an object is there,” or “No, an object isn’t there.” If, for example, one of the parts is rotated or skewed on the line at some point, the proximity sensor won’t detect it. The robot arm typically attempts to pick up a piece, expecting it to be in the same orientation as all of the previous pieces, often with subpar results. This can damage the product being picked up and the robot arm itself.

Designers add machine vision to many robotics applications as a form of guidance in place of simple proximity sensors. With a machine vision-based solution, the designer can send the robot arm the trigger to start the pick and place as well as the rotation and exact coordinates of the piece in question. This reduces missed parts, increases machine efficiency, and, in turn, saves the end user time, money, and headaches. Vision also reads 2D codes on parts to verify that the correct piece is picked up, and it ensures that nuts, bolts, and washers have been added correctly.

There are many other applications that are well-suited to having a machine vision system in place. These range from verifying labels on soda bottles to counting the number of pills that are poured into a bottle as it’s prepared for a pharmacy. Many no-contact measurements that need to be made can be performed with a machine vision system quite easily as well.

Once designers realize that they need to add machine vision to their new designs, the question comes back to which approach works best for them - an off-the-shelf machine vision system (like a smart camera) or a custom design. Cost is a major concern for many machine builders looking at off-the-shelf hardware. Some designers believe that they can save money by going with a custom design over an off-the-shelf solution. Yet, there are several costs associated with a custom design that many designers don’t realize until it’s too late. Understanding these costs up front makes a commercial off-the-shelf product a much more viable option.

One consideration is the time it takes to design a custom system versus the time it takes to implement a system using off-the-shelf parts. Putting together a custom vision sensor and setting up I/O on that sensor to communicate with the rest of the machine is a large task. Years go into the development of many custom designs, and time spent on engineers is expensive. This is often the highest cost associated with custom designs.

There are also hidden costs connected to custom designs that are not found in off-the-shelf hardware. Support is a large hidden cost. Designers not only have to support the operation of the machine built (which would be the case with both custom and off-the-shelf design), but they also have to support the intricacies of a custom design. When something in the custom design breaks, it may be costly to send someone to the site who has any idea what this custom design is doing or how to fix it. With off-the-shelf products like the smart camera, the support burden falls on the vendor of those products. If something happens to the machinery and the system stops functioning, off-the-shelf products offer a safety blanket on the support side.

Smart cameras are examples of commercial off-the-shelf products that can provide machine designers and engineers with the flexibility and power they need to meet the advanced application requirements set before them without having to turn to custom design. There will always be applications where custom design is necessary, but as off-the-shelf measurement and control hardware and software continue to become more powerful and lower in cost, they offer a much more practical solution.

This article was written by Matthew Slaughter, Vision Product Marketing Engineer at National Instruments, Austin, TX. For more information, Click Here