Jan. 28, 2015 —
Tie a long string to a paperweight. Pick up the weight using the string’s far end and move it as fast as you can several feet. The weight will probably swing back and forth in a predictable pattern. Now imagine a variable-speed fan blowing randomly on the weight. How do you move it precisely and rapidly with little or no sway?
That’s the challenge facing crane operators. Fortunately, there’s a solution.
But, getting it right took engineering and time, as well as machine vision technology and proper camera selection, according to Joseph Discenza. He’s president and CEO of SmartCrane. The company’s software allows trained but not-necessarily-experienced crane operators to move loads at maximum rated speeds under normal conditions, thereby boosting productivity.
As for abnormal conditions, examples include an off-center lift, blowing wind, and slippage in mechanical controls. These and other factors can cause crane loads to sway unpredictably. SmartCrane’s technology minimizes the problem.
Machine vision plays a key role in making this happen. The way to stop random sway as reliably as possible, Discenza says, “is to construct a black-and-white pattern of some sort and attach that down near the load and put a camera up near the trolley looking down.”
Tracking the motion of the target yields the info needed to counter externally induced sway. Small moves of the right size and at the right time can get rid of unwanted motion.
That simple explanation hides a lot of engineering and hard-won knowledge. For instance, the design of the target is critical. Cranes operating outdoors can be in full sunlight or deep shadow during the day. At night they work under bright lights.
One key to handling such variable illumination is the target, Discenza says. It needs to be as immune as possible to shadows and also not given to reflective glare. Being insensitive to distance is also important, since the target may be nearer or further way at different times.
The solution that SmartCrane came up with was a checkerboard pattern set at 45 degrees. This is constructed out of a special white background material that reduces reflected glare, with the soft side of a fabric hook and loop fastener used for the black. By finding the target and looking at it, the software can determine its periodic motion and how to correct for it.
Of course, you need to have a camera to see the target. That brings its own demands, particularly since it isn’t sitting inside a lab somewhere. Instead it’s located atop a crane, a large mechanical contraption subject to mechanical stresses and vibration.
This was clear early on when the system was deployed in a big container crane. As the trolley moved along, it jolted over joints, repeatedly rattling the camera and other machine vision components.
“One of the first times we went up to find out why the camera wasn’t working, the lens was gone. I mean, it was powder,” Discenza recalls.
Figuring out how to prevent such a problem is one example of hard-won expertise. Another is the use of longer focal length lenses, which when combined with the right aperture setting can avoid depth-of-field issues.
The result of this experience, engineering and machine vision technology is that loads sway as little as possible while safely getting to their destination faster.