At first sight

There can be no doubt that robots are now an integral and vital cog in the automotive manufacturing industry, but this was not always the case.

Just 25 years ago, only about 6,000 robots were at work in factories in the United States of America, for example. Since 1982, according to the Robotic Industries Association, that number has exploded to more than 170,000. More than a million industrial robots are at work around the world,and the advent of the assembly line robot has transformed manufacturing.

If one single technology has advanced the use of robots within the automotive sector, then it is vision systems. Vision for robots has been around for a long time, but only in the last few years has it has met the main criteria for widespread adoption - low cost and high reliability.

Now, the true cost of implementing robot vision is falling rapidly, while reliability is increasing and ease of use is improving.

Seeing the future

The trend of vision systems becoming a fully integrated feature of robots will continue into the future, and at a faster pace than at present. As vision systems become more robust, more capable, easier to use and less expensive, vision is on the path to being a standard feature in robots.

Vision systems are not new in automotive robotics, but just like the robots themselves, vision is gaining greater penetration because of its reliability and tumbling costs. "I think that the use of vision systems is a function of the integration possibilities with the robots themselves," Rick Youngblood, plant manager of truck trim and chassis at Nissan, US, says. "There are a lot of quality enhancements that are possible using a vision system. One that comes to mind in our plant is the application of mastic and sealant and being able to pinpoint the application area. I think that that is one primary area where we use the vision system.

"There has also been a great deal of enhancement in the last couple of years with laser measuring for body accuracy. At one point [we would] take a unit out of the system, take it to a layout machine at some point, and either manually or later on automatically lay it out. We have got that capability in line and can do it on every unit that we build."

Roberta Nelson Shea, general manager at Pilz Automation, speaks of developments in vision systems that will become available in the future. "I see new vision products, like safe vision and camera systems for robotics. Both are in use by Mercedes-Benz in Germany and sample systems in the US are being tried out."

Robot makers have begun to embed vision into their robots, making it a standard feature. Gary Zywiol of Fanuc agrees: "Fanuc shipped 50 per cent more robots with vision in 2008 than in 2007, and we expect that to continue in 2009. In the last five years, our robotic vision shipments have increased by 400 per cent, so vision is growing faster than the rate of robotic growth."

Vision-ready robots

More robots are vision-ready, which allows users to simply plug in a camera. Integrated vision is therefore a handy feature for end-users new to robotics, who initially might not see the need for vision but could do so as their manufacturing requirements change.

"Throughout the industrial robotics community, the general message is that 3D guided robotic applications are proving themselves both in terms of reliability and return on investment," says David Michael, director of core vision at Cognex.

"In most technological innovations, it's usually very easy to overestimate in the short term and then underestimate in the long term the impact," adds Adil Shafi, president of Shafi, a Michigan-based robotic vision supplier. "3D vision has reached the inflexion point, at which there's rapid and massive adoption."

He goes on to explain that there are only two requirements for a robot to work with vision and to see things randomly in space - you need to be able to talk to the vision sensor, either through a PC or directly to the sensor, and secondly, you need to be able to see that the robot updates its location when it's given some feedback by vision.

Babak Habibi, president of Braintech, a Canada-based company that develops vision-guided robotics software, allies 3D vision with the term visionary robotics. "A good way to understand it is to contrast it to what traditional robotics is," he says. "Robotics has been perfected over many decades to basically provide a machine that is capable of playing back motion in a very repeatable fashion, a very fast fashion and very reliable fashion.

"Robots aren't very good at dealing with change in the environment, and visionary robotics presents cameras and software, mathematics, 3D geometry and processing to create robots that are adaptable to the changes in the environment and what that does is two-fold.

"One, it means that certain applications that were not possible in the past, now are within the realms of possibility. So it opens up all these markets for robots and opportunity to sell more robots, and it reduces the amount and complexity of equipment you need to make a robotic cell operate, which means that the return on investment is improved."

This he explains as a coming-together of a number of key factors: "Some of these factors have to do with general enabling new technologies: cheap and abundant processing power in the form of PCs off the shelf, software development tools that are much more sophisticated, and we have communication protocols and fieldbus that didn't exist previously.

"In terms of vision and robot hardware, we've come a long way as well. We have cameras, visual cameras that are compact, they're fast, and they're controllable. We have LED lighting for machine vision which is much more reliable, about ten times more life than the fluorescent and incandescent predecessors, and we obviously have much cheaper and much more precise and reliable robots. So these have come together to help make visionary robotics a reality today.

Vision and guidance algorithms

"There have also been some really important advances in vision and guidance algorithm development, making it possible to find features on objects reliably. Traditionally we had to use correlation-based pattern matching which relied on light levels, and that was very, sensitive to changes in lighting and appearance. There have been some great advances algorithms to find the 3D position and orientation of parts. Other important advances include the emergence of dedicated tools for vision guidance specifically, as opposed to machine vision. Before, you had to use general-purpose machinery and software, and trying to do 3D guidance with that is very difficult because it's just not meant for that. Now we have actual platforms that are dedicated to this area. We also have higher levels of integration that are emerging between robots and vision, which is very important and is a key aspect of moving forward with 3D vision."

One man at the sharp end of selling robots with vision systems into the automotive manufacturing sector is Kuka Robotics automotive sales director Michael DeWitte. His view is that yes, car makers are using more robotics in their body shops to take care of tasks that they used to do with hard tooling or fixed tooling transfers. He explains that they are working towards manipulation of parts that are loaded to fixtures that are mounted on the robots and then presented to another robot for welding.

"They are starting to look towards better weld quality and tracking of welds, logging the quality of the spot-welding and being able to track that to particular vehicles," he explained. "You are also starting to see a higher amount of material robots - it used to be 70:30 between spot welding and handling, [but now] you are starting to see those percentages equalise. That's driven by the vision systems and the safe technology being developed by robot manufacturers, which allow operators to interface directly with the robots. That allows operators to come into contact with the robots without switching them off."

It is clear that the maturation of vision system technology is opening new avenues for industrial robots. In the future we can expect greater integration between robots and vision to enable companies with heavy reliance on automation to continue leaning out their manufacturing operations as they strive to combat the darkening economic landscape.