Robots ‘See’ Factory’s Future

Is the factory of the future on the industrial horizon? Over the past few years, what used to be considered too far ahead in the “imagine this” category has begun to look more doable. Consider the incorporation of vision technology into advanced robots, particularly in the automotive industry. 

Braintech of Vancouver, British Columbia, is a leading researcher and developer of state-of-the-art vision systems for manufacturing. Recently, the company’s vice president

ADVERTISEMENT

communications, Vincent Taylor, told MHM how his company teamed up with ABB Automation in Brampton, Ontario, to develop and install “seeing” robotic systems for major suppliers to auto-makers in the United States. These systems have been running for several months and are showing major increases in uptime, quality and overall productivity while reducing costs.

For decades, manufacturing engineers have been attempting to increase levels of automation in the factory using computers and the latest material handling technologies, including robotics and systems integration. Progress has been steady but frustratingly slow.

“The difference now,” in terms of robotics and vision, “is it works,” says Dr. David Wright, vice president strategic research at Braintech.

One factory that shows how it works is in Ossian, Indiana. It is a plant of TI Automotive, Warren, Michigan, which manufactures plastic gas tanks for light trucks and cars. The gas tanks are blow molded, machined, assembled and inspected before shipment. The plant employs 500 humans and 100 ABB robots, some of which are now “seeing” robots.

Before installation of the multiple camera 2D vision systems on those robots, they “would blindly do their work,” assisted and supervised by workers. The result was high reject rates. The vision systems now allow the robots to identify “exact positions of critical features including brackets, hole positions and connecting points,” Wright explains.

Expensive fixturing was replaced with vision, and time per operation was reduced to a few seconds from several minutes to identify features and communicate up the line to the other robots for plastic welding and drilling operations.

A further advantage is reduced cost of retooling (which is often in the many thousands of dollars) and readjusting the production lines due to design changes. Now much of the reconfiguring is basically a quick software change for the vision systems. This takes days now rather than several weeks. 

The cost savings in productivity that these new vision-equipped machines offer range between $2,000 and $15, 000 per minute, or output-per-hour gains of 10 times over the original robotic systems, Braintech notes.

One source of such gains is the terrific decrease in the cost of computer power over the past few years. Today’s PCs can easily handle the complex algorithms that account for the robots’ vision capabilities.

While the robot industry in the United States last year reflected the general slump in manufacturing capital goods, reporting about a 32 percent drop in unit orders for the entire year, total robots in the United States working in factories are now around 116,000. That’s less than half of Japan’s number, but well ahead of any other country. With developments in vision and increasing accuracies and flexibility of operation as well, some experts see robotics as a major growth industry.

“The game has changed,” says Taylor. These new vision-equipped robotic systems offer a “way to reduce sharply the cost of manufacturing. They also offer enormous savings in workers’ costs in terms of ergonomics. It’s a fundamental change in manufacturing.”

So, while the world continues to worry about accounting methods and various kinds of swindlers in the business world, there is some remarkably optimistic news about manufacturing on the horizon. First, productivity in the manufacturing sector has continued to grow at surprisingly high rates right through this downturn and stock market decline, according to the Department of Labor. And according to these experts in industrial automation, it will grow even more spectacularly in the near future — starting now in some automotive plants.

George Weimer, contributing editor, weimerg@fleishman.com