About this role
Master's Thesis: Computer Vision for identifying completed manual assembly activities in an industrial environment
Background Manufacturing environments increasingly rely on detailed production data to understand process status, quality, and production progress. Existing production and automation systems can often capture information directly from machines, PLCs, and other technical equipment. Manually performed assembly activities, however, are generally more difficult to automatically observe and register.
Computer Vision and machine learning methods provide an opportunity to analyze image or video data from an assembly station and identify objects, activities, and process states. Such a solution could potentially complement existing production and automation systems by automatically identifying when specific manual assembly activities have been completed.
Purpose The purpose of the thesis is to investigate how a Computer Vision-based solution can be designed to identify completed work activities at a manual assembly station.
The study should be limited to one selected workstation and a defined number of assembly activities related to the assembly of engine components.
The thesis should consist of two main parts:
- Design and evaluate a concept for using image or video data to identify one or more defined assembly activities.
- Evaluate and compare relevant Computer Vision and neural-network-based methods for understanding and performing this analysis.
The evaluation should not only consider theoretical model accuracy, but also the suitability of the approaches for a real industrial environment.
Relevant evaluation criteria may include:
- Detection or classification
- Robustness
- Required amount of training data
- Inference time and computational requirements
- Camera positioning and viewing angle
- Lighting conditions
- Occlusion of objects or components
- Potential integration with existing production and automation systems
A possible overarching research question is: How can Computer Vision be used to automatically identify completed manual assembly activities, and which technical approaches are most suitable for this task in an industrial production environment?
Data Collection Data collection should be limited to a selected assembly station and a defined set of assembly activities. The data collection will occur at Volvo Penta’s Vara Factory.
The work may include:
- Collection of image and/or video data from the selected workstation
- Definition of the assembly activities that should be identified
- Annotation of relevant objects, activities, or process states
- Creation of training, validation, and test datasets
- Observation of variations in how the activities are performed
Based on the characteristics of the selected activities, the student may evaluate methods such as:
- Object detection
- Object tracking
- Pose estimation
- Action recognition
- Temporal video analysis
- Combinations of several Computer Vision methods
The thesis should focus on selecting and motivating suitable technical approaches and empirically evaluating their performance for the selected industrial use case, rather than being limited from the beginning to a specific neural network architecture.
The final outcome should consist of both a technical evaluation and a recommendation for how a Computer Vision component could complement existing production and automation systems.
Relevant Academic Fields Suitable for students in Computer Science, Artificial Intelligence, Machine Learning, Data Science, Automation, Robotics, Mechatronics, Electrical Engineering, or related fields.
Recommended Level Suitable for Master’s thesis work.
Who we are and what we believe in We are committed to shaping the future landscape of efficient, safe, and sustainable transport solutions. Fulfilling our mission creates countless career opportunities for talents across the group’s leading brands and entities.
Volvo Penta, a world-leading supplier of engines and complete drive systems for marine and industrial applications, you will be part of a global and diverse team of highly skilled professionals who works with passion, trust each other and embraces change to stay ahead. We make our customers win.
Job Category: Technology Engineering
Organization: Volvo Penta
Travel Required: No Travel Required