Machine Vision in the Paint Shop for Inspecting primer on the coat

By June 2, 2022November 28th, 2023Industry Use Cases

Quality control is a critical element in the automotive manufacturing industry supply chain with visual part inspection being a heavily relied upon activity in the process. This task has been traditionally carried out by highly experienced and qualified staff. However, the inherent nature of visual inspection and difference in perception between individuals, make defect detection subjective. These inconsistencies have an adverse impact on quality and lower the overall productivity of the plant. One of our clients, a leading car manufacturer, approached us to find a solution to a similar problem. The application uses Machine Vision in the Paint Shop for Inspecting primer on the coat.

Problem Scenario

In the given scenario, one car passed through the painting station every 30 seconds during the primer coat application process at the paint shop. As the robot applicator deployed for the task performed its job, the completion and quality of the primer application was visually inspected by humans. Any defects detected during the inspection were corrected manually by the human inspectors. The main area of concern was the primer missouts that could occur due to blocked nozzles, paint reservoir running empty, and so on. If undetected, these miss outs could result in the paint coat peeling off from that location due to absence of primer.

Primer inspection on a coat

Challenges and Concern

Errors in manual detection of paint miss outs could arise due to the vehicle body moving on the conveyor while being inspected, inadequate access to the area of the panel for visual inspection or sub-optimal lighting. Scrutinizing the surface also becomes an exacting task as the area subject to inspection may exhibit different specular characteristics because of varying shape, size, curvature and material. The planned cycle time of 30 seconds puts latent pressure on inspectors, further aggravating the challenges of visual part inspection. The next step after primer application consists of a baking process, which causes the primer to adhere to the chassis much more firmly. Hence, any omission during human inspection becomes rather costly to rectify.

Also, the process involved about 6 to 9 inspectors inspecting the vehicle, spread in 3 shifts over a
24hour period, and the annual cost to the company for this human labor is quite significant. Being a leading automotive OEM with strong commitments to worker health and safety, another concern was the risk to health of the personnel due to extended exposure to chemicals during primer application.

Solution

As part of hardware component of the Qualitas EagleEye® Platform, a setup with four cameras was installed at the painting station to visually inspect all the regions and magnify the defects. Two cameras were placed fore and aft below the car, and one camera was installed on each side.  For high quality image acquisition, 40 FPS cameras were used at a working distance of 1.5m with a 5mm lens for the rocker panel area, and an 8mm lens for the deadener. The lighting arrangement installed at the bottom eliminated glare and shadows, and provided uniform illumination in accordance with the surface of the material being inspected.

The next step in the process was using Qualitas EagleEye® – a machine vision software to prepare the data for training and processing. A sufficient number of rejected images were added to train the system to identify the defects. The captured set of images were labeled as “OK” or “Not OK” based on the primer coating seen in that image. This annotated data was then used in supervised learning. The system analyzed this data, and trained corresponding models of the defects to be identified. The trained model spotted defects and missouts in real-time and sent signals to the system to generate an audio alert along with screen visuals. The human inspectors then intervened manually and performed the requisite corrective action.

Qualitas AI platform for primer coat inspection

Results

Deployment of the machine vision solution from Qualitas EagleEye® Platform achieved almost 100% accuracy in identifying primer coat defects for a minimum defect size of 5mm, and reduced the inspection cycle time from 65 seconds to less than a second. The AI-based automated inspection solution offered a better reliability quotient than the human based inspection, thus resulting in a potential reduction in costs of remedial action and labor. It also dispelled concerns related to worker health and safety, by minimizing human exposure to chemical agents used in the process.

One Comment

Leave a Reply

Schedule A Demo
close slider


hbspt.forms.create({ region: "na1", portalId: "726123", formId: "bd9f048a-7e04-4512-9179-c50855c961ea" });