The automation of assembly processes involving flexible and deformable components poses a particular challenge due to their high geometric variability. In contrast to rigid components, the shape of these parts strongly depends on external influences and previous handling operations, making reliable detection and positioning difficult. After singulation, the components are often only partially ordered. For automated downstream processing, information such as component type, position, orientation, characteristic shape states, and quality features must therefore be acquired. Modern computer vision techniques and machine learning methods enable the automated extraction of this information from image data and thus constitute a key prerequisite for the implementation of automated assembly processes.

OBJECTIVES:

The objective of this thesis is the development and investigation of a computer vision-based method for the automatic detection and analysis of flexible and deformable components. First, the requirements for the visual acquisition of the components will be analyzed, and suitable image processing and AI-based methods will be researched and evaluated. Subsequently, a method will be developed that enables the following functionalities:

• Identification of the component type
• Determination of the component position and orientation on the support surface
• Detection of visible defects, particularly cracks or other forms of damage
• Transfer of the acquired information to a robotic path-planning system

The developed methods will be evaluated experimentally with regard to detection accuracy, robustness, and real-time capability. The thesis aims to provide a foundation for the vision-based handling of flexible and deformable components.

USEFUL PRIOR KNOWLEDGE:

• Basic knowledge of computer vision
• Interest in AI-based analysis methods
• Strong interest in mechatronic machine components