Background

How can a factory autonomously adapt to constantly changing conditions? The research project "AgiProbot" at the Karlsruhe Institute of Technology (KIT) deals precisely with this question. Remanufacturing is an ideal use case: used motors come back to the factory in an unknown condition should be disassembled as automatically as possible and selected components returned to the production processes. As a subtask of the project, this thesis topic focuses on learning features from motor data with deep learning-based methods. A particular challenge is the use of point clouds as input. In addition to the use of real-world data, a special Blender addon has also been developed to generate synthetic data of various motor instances, with which a large dataset may be created for the needs of deep learning tasks.

Tasks

In this work, firstly a large synthetic dataset will be generated with the provided Blender addon. With the synthetic dataset, different neural network architectures designed for learning on 3D point cloud data could be used for the segmentation task. Not only the segmented results, some important explicit parameters of motors, e.g., the position of bolts, should also be obtained. Possible architecture modifications may also be made for performance improvement. Finally, an end-to-end workflow should be designed to make sure the method works in the real remanufacturing pipeline.

References

[1] Qi, C. et al. “PointNet: Deep Learning on Point Sets for 3D Classification and Segmentation.” CVPR (2017): 77-85.
[2] Wang, Y. et al. “Dynamic Graph CNN for Learning on Point Clouds.” ACM Transactions on Graphics (2019): 1-12.

Conditions

• Subject: computer science, mathematics, electrical engineering, applied physics with good programming skills
• Willingness to familiarize yourself with new topics and enjoy bringing in your own ideas
• Good English speaking and writing skills, ability to work independently and strong analytical skills
• Good understanding of the basics of deep learning, experience of DL projects is a plus

We Offer

• Intensive support and a pleasant working atmosphere in a creative team of motivated scientists
• Possibility of a subsequent job as a research assistant in order to further deepen the knowledge acquired
• Development of joint publications

Contact

Chengzhi Wu
chengzhi.wu@kit.edu