Thesis Code: 24018
Research Area: Connected Systems and Cybersecurity
Description
This thesis investigates the application of deep learning techniques for real-time object detection and tracking using 3D LiDAR point cloud data. The objective is to develop a system capable of detecting and tracking multiple objects in dynamic environments using deep learning architectures. The research focuses on processing point cloud data from a 3D LiDAR sensor to identify, classify, and track various objects such as vehicles, pedestrians, and cyclists. The proposed system will implement recent advances in point cloud processing networks (e.g., PointPillars, PointRCNN) combined with multi-object tracking algorithms to achieve high-performance real-time detection and tracking.
The evaluation will be conducted using both established benchmark datasets and real-world road scenario data collected from actual LiDAR sensors. To validate the system’s accuracy, the results will be compared against camera-based detection methods and ground truth annotations, providing a comprehensive assessment of the LiDAR-based approach’s effectiveness in real-world applications.
Requirements
- Computer Science, Mechatronics, or similar background
- Experience with deep learning frameworks (PyTorch, TensorFlow)
- C++/Python programming languages
- Experience with point cloud processing will be considered a plus
- Linux OS knowledge will be considered a plus
- Proactive mindset, problem-solving oriented
Duration: 6-8 months
Contacts: send a resume with attached the list of exams to federico.princiotto@linksfoundation.com and daniele.brevi@linksfoundation.com