Boosting Automotive Courtesy Light Production with Machine Vision and Robotics
Some automobiles feature courtesy lights that project a brand’s logo or other patterns onto the ground using light. They’re usually affixed to the car door and illuminate when a person enters or exits the car.
Robooptic Systems (Klagenfurt am Wörthersee, Austria), an integrator specializing in optics and robotics, worked on a project to automate assembly and inspection processes for the courtesy lights.
Part of the GAW Group (Graz, Austria), Robooptic Systems developed and installed the automated process for a major manufacturer in the automotive industry. The system automates three out of five steps in the complete process to assemble the projector light module, while other machines handle the remaining steps.
After the manufacturer developed a new courtesy light product, the company hired Robooptic Systems to design and implement an automated process to assemble and inspect the new light. The process of aligning the parts during assembly is more complex with the new light than was the case with the old light, requiring a new machine, according to Robooptic Systems.
Related: Seco Tools Installs Custom Solution to Inspect Indexable Inserts Using in Machining Operations
Three Machine Vision Processes
Robooptic Systems’ machine includes the following steps:
- Getting a part—barrel-shaped plastic container with an integrated collimator lens—out of a carrier tray and adding four dots of glue inside the barrel.
- Putting a mirrored negative with the pattern that will be projected inside the barrel. Robooptic Systems refers to this part as the “slide”.
- Adding a ring of glue to the barrel followed by installing a focus lens to the top of the barrel.
All three processes contain sub-steps such as properly positioning robots and inspecting outputs.
The engineers started the project early in 2023, delivering the finished machine to the customer in December of 2023.
The setup sits in a large enclosure with a SuperTrak conveyor system from ATS Automation Systems (Cambridge, ON, Canada) that moves the barrel assembly from one station to the next one.
To begin Robooptic’s process, trays containing the barrels with the collimator lens are loaded into the machine and four dots of glue are added to the inside of the barrel.
Aligning and Mounting the Mirrored Slide with Pattern
The next series of steps involve not only robotics but also machine vision, says Robooptic Systems.
First, an ABB (Zürich, Switzerland) Scara 910INV robot picks up a mirrored slide and moves it closer to a camera setup. This allows the camera system to produce images and then measure the position of both the barrel and the slide. The camera setup takes a photo of the barrel and another of the mirrored negative, and then the system merges the images together.
The camera system also inspects both parts and ensures the glue was dispensed correctly.
The machine vision system at this station includes components from Opto Engineering (Mantua, Italy): an ITA120 monochrome camera from with a RT-TCL0450-FU optic and an integrated coaxial white LED. The system also includes a small bar light with diffuser to light up the slide from behind.
Related: Next-Generation Robots will Rely on Integrating AI and Motion Control
Using the measurements, the robot puts the mirrored slide into the barrel. Meanwhile, the overall system turns on a UV light to cure the glue.
The assembly then moves to the next station where two linear axes in combination with a CPX motion controller dispense a ring of glue onto the barrel.
Using a Telecentric Lens to Inspect the Glue Application
After the glue ring is applied, a machine vision system inspects the glue application. This is an important step because spots of glue or glue overflow on the barrel are big issues for the machine. For example, glue could end up on the robot’s gripper, creating all sorts of issues.
This station uses an ITA81 colored camera alongside a TC12024 telecentric lens—both from Opto Engineering. The lenses are "designed to accept only optical rays that are parallel to the optical axis. Since telecentric lenses only accept parallel optical rays, the magnification of a telecentric lens remains constant over a range of working distances and is independent of the object’s location. Therefore, an object is always displayed at the same size at different working distances,” says Beatrice Danese, Product Manager at Opto Engineering.
Related: Telecentric Lenses with Variable Working Distance
To illuminate the part, a blue coaxial light (OE LT2QOG025-00-X-B-24V) is mounted underneath the optic. The blue light illuminates the glue with a fluorescent glow, which helps with the inspection process, but isn’t far enough into the UV spectrum to cure the glue.
Adding a Focal Lens to the Assembly
In the next step in the process, a robot aligns and then mounts a focal lens on to the barrel assembly. There were three identical stations, each consisting of a PI H811.I2 Hexapod robot (Physik Instrumente; Karlsruhe, Germany) because it is the process that takes the longest.
Each station also included an ITA89 monochrome camera and an 8 mm EN8MPL0818 fixed focal length lens—both from Opto Engineering. The setup also included a 3 W LED.
Several steps occur at these stations. The barrel assembly is picked up by an axis and is attached to a circuit board with an integrated LED by using prongs to hold the part in place.
Related: Machine Vision System Detects Antenna Trace Defects on Automotive Glass
Next, another axis picks up the focal lens and puts it inside the robot’s gripper. The robot then aligns the lens to the barrel/circuit board and then mounts the lens.
Next the glue is cured with UV light, and the part is put back into the carrier.
Final Machine Vision Inspection
At the last station, a final inspection occurs. An ABB Scara 910INV robot picks an assembly out of the carrier and moves it in front of a Fal.6-4-857724-00-375+ UV light from Falcon Illumination (Penang, Malaysia). A 12 MPixel camera from Baumer (Frauenfeld, Switzerland) produces images, allowing the vision system to determine if the part is OK or NOK (not OK). The system is primarily looking for cracks.
Based on the results of the inspection, the robot moves the part into the correct tray for unloading from the machine.
Software Programming Challenges
To program the software for each step in the machine vision part of the process, engineers used Opto Engineering’s FabImage Studio Professional Software, which is flow-based software for machine vision tasks.
However, the engineers used custom programming for some functions because the system includes so many steps involving machine vision, motion, and robotics. For example, the alignment algorithms were complicated. In the step where the robot puts the mirrored piece into the barrel, it was difficult for the engineers to get the parts to line up exactly, Robooptic Systems says.
Perfect alignment is crucial in this application. If the parts aren’t aligned correctly, the image projected on the ground will be fuzzy.
Related: Robotic System Loads Bobbins in Textile Manufacturing Process
To complicate matters more, there were multiple patterns—about 15—in the project, each requiring different alignment data. Robooptic Systems says the team spent several months determining the ideal coordinates to align each type of patterned slide with the barrel. The engineers created a recipe-type system where they recorded all of the alignment data in the software, according to Robooptic Systems.
In the end, however, they were successful. The entire process takes less than seven seconds per part and less than a few percent of the parts produced are classified as NOK during the final inspection.
About the Author
Linda Wilson
Editor in Chief
Linda Wilson joined the team at Vision Systems Design in 2022. She has more than 25 years of experience in B2B publishing and has written for numerous publications, including Modern Healthcare, InformationWeek, Computerworld, Health Data Management, and many others. Before joining VSD, she was the senior editor at Medical Laboratory Observer, a sister publication to VSD.