Auto-Labeling System with Agentic AI for Perception in Auton-omous Driving
- Type:Master's thesis
- Date:Immediately
- Supervisor:
Background:
In autonomous driving, high-qual-ity labeled data is essential for training perception systems, ena-bling them to accurately detect and track objects in complex environ-ments. However, manual annota-tion is a highly labor-intensive, time-consuming, and expensive process, particularly when dealing with large-scale multi-modal datasets such as LiDAR and RGB, as well as collaborative perception datasets involving multiple perspectives. To address these challenges, automated auto-labeling sys-tems have been developed to improve efficiency and reduce dependency on human annotators. Recent breakthroughs in Large Language Models (LLMs) and Vision-Language Models (VLMs), alongside Agentic AI frameworks, offer transformative potential.This thesis aims to design and im-plement an innovative auto-labeling system that harnesses Agentic AI to advance multi-modal per-ception in autonomous driving. The focus will be on improving annotation accuracy, robustness to challenging conditions, and scalability across LiDAR and RGB datasets.
Your Tasks:
- Analysis of SOTA in Auto-Labeling of Multi-Modal Datset
- Build a system to annotate multi-modal data (RGB, Thermal, LiDAR) using LLMs and VLMs for contextual understanding and validation.
- Validation with Real and Simulated Data to assess performance.
- Benchmark the system’s accuracy and efficiency against existing SOTA methods.
Your Profile:
- Strong background in machine learning and computer vision.
- Experience with multi-modal sensor data processing (e.g., camera images, LiDAR point clouds, GPS time-series).
- Knowledge of deep learning frameworks (e.g., PyTorch or Tensorflow) and practical experi-ence of LLM/VLM
- Ability to work independently and tackle complex, open-ended research problems.
Start date: Immediately
Duration: As per the applicable examination regulations.
If you are interested or have any questions regarding this thesis position, feel free to contact.