Thesis supervisor: László Gulyás
Location of studies (in Hungarian): Faculty of Informatics, Eötvös Loránd University Abbreviation of location of studies: ELTE
Description of the research topic:
Environment and object detection is essential for driverless systems. However, the current single sensor detection mode is no longer sufficient in complex and changing environments. Therefore, this thesis will be concerned with the combination of camera input with light detection and ranging (LiDAR) or Radar to build 3D models of the environment and possible objects (dynamic or static). The major benefit of this sensor fusion leads to a better, more accurate object representation in the different use cases using also projective mapping techniques. With a semantic interpretation of the scanned environment and events a detection framework multi adaptable with high real-time capacity and robustness can be achieved.
The fusion of LiDAR, optical cameras and other sensors will provide high precision 2D and depth perception with 3D point clouds registered to the images of optical cameras. The acquired data will give us the possibility to perform a partial recognition of the environment, structures and objects that can be reinforced with fusion of different image sources from the same or different vehicles or other obstacles in the driverless systems.
This research work will try to improve the 2D/3D representation accuracy and the robustness of the proposed framework by combining the benefits of deep-learning networks and traditional geometric based methods. The research topic covers a wide range of fields mainly robot vision and 3D recognition for the environment comprehension, robot vision-based communication and sensor network for fusion data and existing probability reinforcement and of course autonomous driving.
Keywords: robot vision, robot vision-based communication, autonomous driving, sensor network, 3D recognition, simultaneous localization and mapping (SLAM), graph theory.
Required language skills: English Recommended language skills (in Hungarian): Hungarian Further requirements: SW: python (Numpy, Pandas, Matplotlib), machine learning: TensorFlow, SKLearn, Pytorch, Keras, optional system development: Spark
Proactive student with good communication skills who enjoys working in a team and is enthusiastic about data.
Login
