3D vision automates sealant application in the aviation industry

The aviation industry places the highest demands on quality, process reliability, and traceability. At the same time, pressure is increasing to automate complex manufacturing steps that have previously been carried out manually. This is precisely where a robot-based automation solution from AKEOPLUS comes in: The French company has developed a robot cell that automates the application of a two-component sealant to painted large metal parts. Using the MVTec HALCON machine vision software, the system detects the position of the component, calculates the necessary corrections, and thus enables precise sealant application with an accuracy of ±0.2 mm - across a large working area.

In aviation manufacturing, even the smallest deviations determine the quality of a component. Naturally, this also applies to processes in which sealants are applied to complex geometries. The sealant bead must be complete, uniform, and positioned in the correct place. Gaps, misalignment, or uneven beads can impair the sealing function. Until now, this process could not be automated. Implementation was too demanding because the components vary in size and come in many variants.

Nevertheless, a company from the aviation industry wanted to automate the previously manual process. Specifically, this involved applying a primer and a two-component sealant to bearings made of bronze or stainless steel that are installed in painted metal parts. This environment is unpleasant for employees, not least because the materials used have a strong odor and handling some of the substances can be harmful to their health.

AKEOPLUS was commissioned to implement the solution. The company was founded in 2006 and is based in France, between Geneva and Lyon. AKEOPLUS specializes in robotics and develops customized solutions that combine robot systems, vision sensors, sophisticated algorithms, and artificial intelligence. "This application was not simply about having a robot follow a path. We had to automate a process that had previously been strongly shaped by the experience and care of the employees. The robot cell must handle different components and application areas, detect the actual position of the bearings, and apply the sealant reproducibly," explains Maxime Motisi of AKEOPLUS, an engineer specializing in robot-assisted dispensing applications.

Large working area - manageable thanks to machine vision
Before automation, the process was carried out manually. An operator first had to prepare a batch of sealants. From that point on, there was only a short time window: after around two hours, the material hardened and could no longer be used. The process was physically and technically demanding as well as time critical. In addition, an invisible primer first had to be applied with a brush to all designated areas. This step was labor-intensive and varied depending on the operator.

The objectives of the automation were clearly defined. The entire process was to be automated as far as possible: from the provision of the fluids and scanning the component to applying the primer and sealant. At the same time, the system was to monitor the material flow, detect bubbles or irregularities, and thereby prevent defects in the sealant bead. Because the process runs autonomously, employees could be reassigned away from this activity, which was not ideal from a health perspective. The automation was also intended to make the process more robust against quality fluctuations.

"Flexibility was particularly important to the customer. The system needed to be able to process many different components. The geometries and configurations therefore differ significantly. The robot's workspace is around 2,000 mm x 2,000 mm x 2,000 mm. The challenge is to enable the robot to reliably detect the relevant areas within this large workspace and adapt its movements to the actual position of the component," explains Adrien Queyroi. In addition, the components are not always located in exactly the same position within the robot's workspace. For this reason, a purely mechanical solution for precisely positioning the component was not an option.

A purely mechanical solution was not an option. The positional deviations were too large. In addition, the end customer wanted to reduce the handling of the parts as much as possible. Mechanical probing would have meant additional process steps, more contact with the component, and potentially longer cycle times. It was therefore clear to AKEOPLUS: The application could only be implemented with the help of machine vision. AKEOPLUS selected MVTec HALCON, the standard software for machine vision, as the central component.

Entire workflow is automated
The process begins with a setup phase. During this phase, consumables such as tools and nozzles for the primer and sealant are installed. Initial manual cycles are then performed to dispense a small amount of material. This ensures that the entire material path and all lines are correctly filled and ready for use. The component is then moved into the working area. A positioner roughly aligns the component and moves it close to the target position in front of the robot. The robot then starts the scanning phase. Equipped with a 3D profilometer, it scans the predefined areas of the component. These areas include the bronze bearings around which the primer and sealant will later be applied. The acquired 3D data forms the basis for determining the actual position of the bearings and adjusting the programmed paths accordingly. After the scanning phase, the robot changes tools and starts applying the primer. To do this, it picks up a swab tool, takes primer from the station, and applies it to the designated sealing areas. The robot then changes tools again and begins applying the sealant. It follows the programmed bead paths. The speed is adjusted according to the desired bead thickness: faster for thinner beads, slower for thicker beads.

After the sealant has been applied, the component is fully processed and can be removed from the cell. While the robot "waits" for the next component, it moves to a flushing position. There, sealant is flushed from the mixing system so that the material does not polymerize in the mixer.

Special challenge: machine vision in three-dimensional space
Implementation was demanding because several technologies had to work together precisely. These include a six-axis robot, a 3D profilometer, a laser fork for tool repositioning, a dispensing system for primer and two-component sealant, a swab tool for applying the primer, nozzles for applying the sealant, a component positioner, process monitoring sensors, and a control system with integrated algorithms.

"Each of these components has its own accuracy limits. The robot, the laser fork, the 3D profilometer, the hand-eye calibration, and the algorithms each introduce their own tolerances into the overall system. One task was therefore to control and combine these influences so that an overall accuracy of ±0.2 mm is ultimately achieved for positioning the sealant bead," explains Adrien Queyroi. The key challenge was locating the component and determining its position in space. To do this, the six degrees of freedom of each bearing must be determined. However, these bearings differ in shape and dimensions. At the same time, the required accuracy of the sealant application is very high. Machine vision may therefore introduce only a minimal error into the overall process. In addition, both the customer's large component and the bearings used are subject to manufacturing and assembly deviations. These deviations are greater than the permissible tolerance for the sealant application. The placement of the component in the system is also not completely reproducible. Although the component is pre-positioned in a positioner, this pre-positioning only brings it to within a few centimeters of the target position. This is not sufficient for the actual sealant application.

"The special challenge was therefore to derive a precise robot path from a variable component. This is exactly where 3D machine vision demonstrates its strength. It compensates for manufacturing, assembly, and operator influences and makes the robot-assisted process possible in the first place," says Maxime Motisi.

HALCON as the central software component
AKEOPLUS uses MVTec HALCON for machine vision. The standard machine vision software performs several central tasks in the application. It processes the data supplied by the 3D profilometer, supports the conversion of X, Y, and Z images into point clouds, combines 2D and 3D operators, and uses CAD data and 3D object models.

"MVTec HALCON offers many functions that are crucial for robot-assisted applications using 3D data. These include robust 3D methods, 3D objects models, the combination of 2D and 3D processing, and the ability to efficiently integrate complex machine vision logic into industrial processes. Especially for applications with high accuracy requirements, this combination is a decisive advantage," explains Jan Gärtner, Product Manager HALCON at MVTec.

For AKEOPLUS, the range of functions was not the only decisive factor. The speed of development and integration also played a major role. The workflows were developed entirely using the HALCON library and MVTec's integrated development environment HDevelop. The scripts are then called seamlessly by the AKEOPLUS framework "AkeoSpine" via HDevEngine. This eliminates additional dedicated engineering work for the transition to production operation.

Another advantage lies in operator guidance. When teaching new bearings or new subprograms, the end user follows a guided workflow. This wizard was implemented using HALCON's GUI, windowing, and plotting functions. It guides the operator step by step through the configuration of the Surface Model and the required parameters. During online processing, the configured subprogram is then called to locate the actual bearing based on the actual 3D scan. Finally, HALCON's hardware independence was also a criterion for AKEOPLUS, as each robot cell must be adapted to the specific customer environment. Extensive documentation, examples, tutorials, and Solution Guides also facilitate development.

"For us, it is important that vision engineers can work very autonomously. With HDevelop, the HALCON library, and the visualization options, we can develop, test, debug, and integrate the application into the overall process. This reduces interfaces and accelerates implementation," says Maxime Motisi.

Result: uniform beads and a technological showcase
The robot cell has been in operation since spring 2026. Shortly after start-up, it already impressed the customer. The applied sealant beads are uniform, thin, and consistent. The system thus achieves one of its central objectives: it makes a demanding manual process more reproducible while reducing the burden on employees.

For the customer, the cell is more than a single automation solution. It is a technological showcase. It demonstrates that even complex processes in aviation manufacturing that have previously been shaped by manual craftsmanship can be automated when robotics, dispensing technology, and machine vision work together correctly.

"Many processes that are still close to manual craftsmanship today will be automated in the coming years. The combination of machine vision, AI, and robotics will enable highly adaptive, intelligent, and autonomous systems. The cell developed here demonstrates very concretely what is already possible today," says Maxime Motisi.

About MVTec Software GmbH
MVTec is a leading manufacturer of standard software for machine vision. MVTec products are used in a wide range of industries, such as semiconductor and electronics manufacturing, battery production, agriculture and food, as well as logistics. They enable applications like surface inspection, optical quality control, robot guidance, identification, measurement, classification, and more. By providing modern technologies such as 3D vision, deep learning, and embedded vision, software by MVTec also enables new automation solutions for the Industrial Internet of Things aka Industry 4.0. With locations in Germany, the USA, France, Benelux, Spain, China, Taiwan and South Korea as well as an established network of international distributors, MVTec is represented in more than 35 countries worldwide. www.mvtec.com