Takeshi Honda: In the automated paint inspection process for the Nissan Ariya at the Tochigi plant, there are inspections conducted using the auto inspection and auto measurement. For the auto inspection, after the integrated painting of the body and bumper is completed, there is a finishing phase in which the painted surface is checked for defects such as dust, blisters and indentations (defects that form tiny crater-like holes due to foreign particles adhering to the vehicle body). Previously, skilled inspectors have been checking the paint surfaces with their eyes and feeling them with their hands, now the robots automatically inspect the finished paint surfaces.
Next is the auto measurement. After checking the appearance, the color, thickness (the thickness of the paint film after it has dried can affect the durability of the coating) and distinctness (the level of finish is checked by measuring how clearly the surrounding scenery is reflected) of the coating are checked. Previously, these checks were performed by inspectors, but now they are all automated by robots.

Daisuke Iwahashi: The particles of dust and blisters that are considered defects during paint inspection are microscopic in size, measuring only 0.3 mm in diameter. It is a daunting task to visually or manually identify them, and it requires special training and qualifications to be in charge of this work. So, there were concerns about whether a robot equipped with only a camera would be able to perform the inspections that skilled inspectors perform with their eyes as well as their fingertips.
On the other hand, in the measurement process, inspectors used to manually change their positions to check the color, film thickness, and glossiness of all painted surfaces of the vehicle. This physical burden was quite significant, so we think that automation would contribute to reducing the workload on inspectors.

Honda: The most difficult part during the consideration of implementing the auto inspection was the quantification of dust and blisters. Even a 0.3 mm particle of dust or blister can have a variety of shapes, such as sharp or smooth, and people perceive them differently depending on their shapes. If it is dust or blisters that a customer is concerned about, even if it is smaller than 0.3 mm, the inspector will determine that it is a defect. The important thing is to match the way people in the gemba detect the defects with the way the robot does its detecting.

Iwahashi: This is precisely the know-how that the inspectors have developed. They do not measure the size of each particle of dust or blister with a measuring device, but detect them with their eyes and their fingertips. This time, we needed to put that know-how into quantitative figures.

Honda: First, we classified the dust and blisters into different shapes with the gemba staff, and then analyzed and determined the values while looking at them with the inspectors at the gemba.

Iwahashi: The next issue after quantification is whether or not the robot can check for dust or blisters that should be determined as defects and not miss them. If the level of defects has to be checked again by a human, there is no point in automating the process. I have set up several pieces of automated equipment in my career, but this one was particularly difficult. The system uses zebra lighting (lighting with stripes of white and black areas) on the body and bumpers, and the reflected images are captured by a camera for inspection, but depending on the way the light is shone and the angle of the camera, there may be dust or blisters that cannot be detected. We had to find a degree of brightness and angle that can pick out the dust and blisters of all shapes and sizes. Also, dust or blisters that could be detected on a flat surface sometimes cannot be detected on an uneven surface. The level of difficulty also differed between horizontal and vertical surfaces.

Honda: We divided the body and bumpers into 600 sections and checked each section one by one to make sure we did not miss any dust or blisters. The color of the paint also makes a difference in detection. With dark colors, the image may be too dark to identify, but if the system is matched to a dark color, the light color will in turn be difficult to identify.

Iwahashi: Human eyesight can do the inspection equally well in white or black, but with a camera, it is different. For example, white is more difficult to detect because the lighting gets reflected and becomes visual ‘noise.’ The way the robots are moved is important for inspecting at the proper camera angle, and since the robots with cameras and lighting attached to the tips of their arms were moving simultaneously, it was difficult to adjust them so that they would not collide with each other. Of course, we checked this in advance with simulations, but we couldn’t get it completely right without seeing the actual movement, so we repeated trial-and-error.

Honda: The most challenging aspect of introducing the automated measuring device was the risk of damaging the body and bumper by the robot making contact with the inspection machine while measuring the film thickness and color.

Iwahashi: One particularly challenging aspect was the thickness inspection. We had to use a thin rod-like probe to make contact with the body and take measurements, but if it is pressed too hard, the body may dent. It was difficult to reproduce the delicate hand movements of skilled workers with robots.

Honda: Defects detected by the robot are displayed on a smartphone worn on the inspector’s arm. They check the defect and corrects it if necessary.

Iwahashi： It is important to teach robots to perform human movements correctly because even though they are robots, their movements are based on human movements. As long as the robot is programmed with the correct movements, it will continue to produce the right products all the time. However, if a robot is taught incorrectly, because it cannot judge whether something is good or bad, it will continue to produce poor-quality products. The automated paint inspection process is the culmination of all the experience accumulated by Nissan’s skilled workers.

Honda： We worked with the gemba staff to make the detection level match that of the gemba.

Iwahashi: This is also the result of repeating the trial-and-error process. In the Tochigi Plant, we have taken the lead among all Nissan plants in implementing the introduction of new technological equipment, and the experience of successfully implementing the equipment has contributed to our motivation.

Honda: I want to continue maintaining the attitude of continuously challenging myself as a member of Nissan in the future.

At Nissan Intelligent Factory we’re building the future of mobility
https://youtu.be/ooFrWpxztOk?si=LdDbxjRE5Mkc10QJ