Visual replaces manual
The Harley-Davidson V-Rod is one of the hottest and most sought after motorcycles introduced in recent memory. Sporting a liquid cooled, 115-hp, 1130 cc, 608 V-Twin Revolution engine (developed in a joint venture with Porsche), a hydroformed perimeter frame, a radiator shroud with twin vortex air scoops, and triple disc brakes, the V-Rod is not only a bike powerfully outfitted, but a beauty to ...
This article contains an expanded verion.
Click here to read the expanded Voice of Experience for February 2004.
The Harley-Davidson V-Rod is one of the hottest and most sought after motorcycles introduced in recent memory. Sporting a liquid cooled, 115-hp, 1130 cc, 608 V-Twin Revolution engine (developed in a joint venture with Porsche), a hydroformed perimeter frame, a radiator shroud with twin vortex air scoops, and triple disc brakes, the V-Rod is not only a bike powerfully outfitted, but a beauty to behold.
Like any well-crafted machine, it did not appear on the scene as a fait accompli . The engine for the V-Rod was a new product for Harley-Davidson. During the process of increasing production efficiency for the engine, the engineers replaced a manual inspection process with one driven by a vision system.
Working with system integrator Bachelor Controls and Power Motion Sales, a DVT distributor company, Harley-Davidson addressed the initial inspection problem by attaching a DVT camera to a servo-controlled linear actuator connected to a pneumatic slide capable of extending to the three different positions in which the snap rings are located in the transmission block. This, however, was just the first step in developing a vision system-based inspection, because the working area was too tight for the 60 x 30mm form size of the camera. To solve this problem, a mirror was affixed at a static position on the pneumatic slide between the press and the engine block to allow the camera to capture images. Although this addressed bearing and balance shaft inspections and one of the snap rings, the position of the other two snap rings created lighting issues for the vision system—software filters with the vision software compensated for the lighting conditions. Finally, to ensure repeatability of the inspection process, a reference point on the snap rings had to be identified as a reference point for camera measurements. The snap rings, similar in shape to a horseshoe, sport two holes in the ring near the gap. Initially, the holes in the ring appeared to be the obvious feature to use in establishing a fixed location, but that didn't pan out for each snap ring because of their location within the transmission. Other options had to be explored.
Control Engineering editorial director David Greenfield spoke with Harley-Davidson manufacturing engineer Dan Bruyn to learn more about the complex, yet "straightforward" decision-making involved in establishing the V-Rod transmission inspection process.
Q How typical was this project for you in terms of engineering decision-making?
On a macro level, the decision-making was very straightforward. We had a quality issue where there was potential for a non-conforming material to be produced. So we brainstormed and discussed possible options, went through the whole decision-making process, and determined how much we could spend based on rate of return. On the micro side, the complexities of getting all the details to work—as far as vision systems talking to PLCs and the operator interface, and getting all of those interfaces connected and working together—that part of the project was more unique. The decision-making process surrounding this project was not new, but since the technology is new, you have to have a different knowledge base to make sure you know what you're doing to be successful.
When I think back about my experience with vision systems, if somebody suggested putting in a vision system a few years ago, everyone would have been scared of the idea because it would be a $50,000 or $100,000 commitment. Therefore, vision systems weren't generally the go-to solution. I'd still say they're not always the go-to solution even today because in some cases it's a lot easier to design and build a fixture quickly, especially for quick inspections. With vision technology becoming easier to use and more affordable, that is changing. You can install a very capable vision system now for about $2,000 in a reasonable time frame. Most fixtures and tools, however, especially if you have anything custom made, are going to be relatively expensive. And a vision system is a better deal long-term because it can be reused on another project, whereas a custom made fixture—a hard gauge or a tool designed for a specific measurement or inspection—is only good for the project for which it was built.
Q Which of the two issues that had to be addressed—camera position and lighting—were the most difficult to engineer?
They were both equally difficult because neither was straightforward and both posed individual challenges that myself and the integrator had never run across before. On the positioning, we had three inspections to do, all at different locations. To reduce the complexity, we wanted to use one camera. To make things even more challenging, the area we're inspecting is not out in the open or easily accessible. Other automated devices were occupying the same area. We have a press installing a bearing that has to be retracted to bring the camera in for the inspection. Then we have to move the camera out of the way so the press can come in and install another bearing and snap ring combination. We handle this with a couple of different motion devices: the camera is moved in and out on a linear actuator and a servo motor is used to focus the camera for each one of the three different locations for the inspections.
The lighting was unique because we used two different kinds. We had three snap rings, two of which were installed behind bearings. So we used a ring light to reflect off the bearing because the light was in line and it worked very well. But in vision applications it's better to have a back light; so for the second snap ring, since there was no bearing behind the snap ring, we used a second light—ultraviolet—to reflect off the transmission case, which backlit the second snap ring. This process wasn't so straightforward either, because we were trying to get away with using just one light, but it was determined that we would get much more repeatable results using this method since we had two different inspections going on.
Q What was it about the holes in the snap rings that made them unusable as the reference point for establishing position?
The entire hole could not always be seen through the vision camera. The lettering on the bearing would confuse the system, and you couldn't always get a sharp contrast in the hole, or you could only see half the hole because the other half was buried in the snap ring groove. So the gap in the snap rings appeared as a robust feature that we could look at time and time again. Because there was almost always a bearing behind the snap ring and, using the gap to focus on, we could always get a good picture of the snap ring to see if it was in place.