Modular Context-Aware Motion Prediction by leveraging End-to-End Driving Stacks

Institute

Lehrstuhl für Fahrzeugtechnik

Type

Master's Thesis /

Content

experimental / theoretical /  

Description

The current state of autonomous driving is split between modular and end-to-end software stacks. Modular systems offer explainability and clearly defined components such as detection, tracking, motion prediction, and planning. In contrast, end-to-end systems treat driving as a single learnable task, enabling strong data-driven context understanding. However, context is hard to capture in purely modular approaches, even though it is crucial for robust motion prediction in real-world driving.

Therefore, this work focuses on using knowledge distillation from end-to-end autonomous driving systems to enhance a modular motion prediction model. The approach should build on Uni-AD or a similar end-to-end stack and transfer its learned contextual knowledge to a modular prediction model counterpart through fine-tuning.

Possibility for publication in case of excellent work.

Requirements

Requirements:

• Very good programming skills in Python.

• High personal motivation and independent working style.

• Very good language proficiency in German, English or French.

Software

Python

Tags

FTM Studienarbeit, FTM AV, FTM AV Perception, FTM Stratil, FTM Informatik, FTM IDP

Possible start

sofort

Contact

Loïc Stratil, M.Sc.
Room: MW 3508
Phone: +49.89.289.15898
loic.stratiltum.de

Announcement