Thesis: GenAI-Enhanced Multimodal Sensor Fusion in Autonomous Driving
Core
Investigate how generative AI can improve multimodal sensor fusion to make 3D object detection more robust under adverse conditions and sensor failures in autonomous driving.
Role type
Master's thesis student (research)
Builds
Generative sensor-completion mechanisms and robust 3D object detection systems
Domain
Autonomous driving, multimodal perception, generative AI
Deliverable
research
Required skills
Python programming, deep-learning frameworks (PyTorch or TensorFlow), multimodal data handling, experimental evaluation
Preferred skills
Generative AI, sensor fusion, robust perception under challenging conditions
Technologies
LiDAR, camera, RADAR, GPU compute resources
Responsibilities
Implement a generative sensor-completion mechanism for degraded or missing sensor information; Integrate the completion mechanism with a pretrained bird's-eye-view object detector; Train and evaluate the system using synchronized LiDAR, camera and RADAR data; Test the approach under modality-dropout and simulated adverse-weather conditions; Measure detection robustness and graceful degradation when sensor information is unavailable; Analyse the results and document findings in a scientific report.
