Bachelor and Master Theses

To apply for conducting this thesis, please contact the thesis supervisor(s).
Title: Synthetic data generation using Unity
Subject: Robotics
Level: Advanced
Description:

 

 

Performance and generalization of intelligent systems place considerable demands on the quantity of training data. Providing sufficient data for training an intelligent system is a time-consuming and costly process. The data labeling process, which requires human supervision, is one of the most difficult aspects of producing a training dataset.  

In this thesis, we will study constructing a data synthesizer using Unity to alleviate data labeling suffering. This synthesizer is designed to generate labeled data without human supervision automatically. The planned data are RGB images and Lidar Point-Cloud, and the intended environment is the construction site.  

We can summarize the objectives of the thesis as follows:  

  1. Develop a data synthesizer using Unity to capture an autonomous driving dataset  

  1. Collect a synthetic dataset with labels for training autonomous driving algorithms.  

  1. Train a model using the collected dataset 

  1. Evaluate the performance of the trained model 

Start date:
End date:
Prerequisites:
IDT supervisors: Ali Zoljodi
Examiner: Masoud Daneshtalab
Comments:
Company contact: