Research project: "SprayOne"

Research area: Automation, automated programming, painting
Supported by: Federal Ministry for Economic Affairs and Energy
In collaboration with: Venjakob Mechanical Engineering & Co. KG
Start of the project: June 2014
End of project: December 2016


The “SprayOne” research project aims to develop a self-optimizing system that paints 2.5D pieces of furniture in a continuous process. 2.5D components have a flat base body with complex geometries, such as outer boundaries and incisions with rectangular but also oblique edges.

The painting is carried out in a division of labor between an industrial robot that paints the complex geometries, while the horizontal surfaces are coated by a subsequent automatic surface spraying machine. The type and position of the components on the conveyor belt are determined based on measurement data from an upstream sensor. To keep the calculation time at runtime as short as possible and to enable human intervention, the robot paths for a component must be generated automatically in preprocessing based on the CAD or, if not available, on scan data. After converting the data into a grayscale image, the component edges are determined using the flat basic geometry. However, an entire path is not calculated, but individual parameterized paths (tasks) are created that can be painted independently of each other and in different directions.

During the run, the system should identify the components, recognize their position and orientation on the conveyor belt, and independently calculate the optimal path across the entire conveyor belt with the arbitrarily positioned components. The previously created path segments of the components must be adapted to the position and placed in an optimal order for the current condition.
There are some special features here:


  • The robot's axis angles and arm lengths are limited. Therefore, it is not guaranteed that all paths calculated in the task space are executable. The problem must therefore be considered in the joint space.
  • As the conveyor belt moves, the paths calculated in preprocessing change continuously. Depending on the start time, the robot has to carry out different movements for a path that remains the same in the task space.
  • The objects to be painted are only in the robot's workspace for a certain time. Previous approaches to task optimization take into account static problems that are located entirely within the robot's workspace. If the planning were carried out using these known approaches, it could be that painting an edge would be planned at a time when it is no longer in the robot's workspace. Therefore, this aspect must be taken into account in the newly developed optimization approach.


Contact person at the institute: Dipl.-Ing. Denise Klose