Autonomous Driving Systems with Large Language Models : A Comparative Study of Interpretability and Motion Planning
Year: 2024
Type: dissertation
Abstract: In this master’s thesis, we investigate the integration of large language models into autonomous driving systems, with a particular emphasis on their potential to enhance interpretability, decision-making, and planning capabilities. We implement both data-driven and knowledge-driven models within the CARLA simulator across diverse scenarios, focusing specifically on the TransFuser and LMDrive frameworks. This study provides a comparative analysis of these models utilizing a range of metrics. The results indicate that while LMDrive exhibits certain limitations in motion planning, it demonstrates significant competence in interpretability, particularly in recognizing traffic light signals and detecting bumpy road conditions.
Author: Yan, Shengheng
Language: English
Subfield: Automotive Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7036414252