Our Mechatronics team has a thesis topic opportunity on offer!

What it is about:
A robot can view objects from different viewpoints, leading to large amounts of training data of the same object(s). To label objects (for later training of AI models) requires a time-consuming manual annotation process.

This project uses a robotic arm, on which algorithms are developed/implemented that automatically annotate all perspectives of an object based on a single manual annotation.
This includes consideration of:
- The robustness of the automatic labeling against robot position inaccuracies.
- A foundation of a Graphical User Interface (GUI).

Upon successful implementation of this project, researchers and engineers can more quickly and efficiently generate training datasets for computer vision algorithms in robotic applications.

Tasks for you:
- Analyze/implement existing algorithms, as well as objectively evaluate the accuracy of these algorithms
- Making implementations more robust, to improve error margins
- Basic implementation of a GUI

Knowledge:
- Interest in robotics and machine learning/computer vision.
Software Skills:
- Experience with Python, (C/C++).

Hardware skills:
- Basic knowledge of robotic systems and cameras (RGB, depth cameras).
- Interest in robot control and positioning.

People skills:
- Able to perform and analyze experiments independently.
- Collaborate with researchers and engineers at Sirris.
- Document results clearly.