Improving Egovehicle Control in Deep Learning training Pipeline for Autonomous Cars

This is Syamimi’s MSc work. We continue our exploration in self-driving car technology.

Game-Theoretic Integration in Imitation Learning for Safe Lane-Changing

Safe driving remains a major challenge for autonomous vehicles, particularly in complex and safety-critical scenarios. While imitation learning (IL) has shown promising driving performance, IL-based models often struggle with robustness when faced with rare or high-risk events. In this work, we study the integration of game theory within the IL pipeline.

People

Data/Codes

Publications

Media

Vehicles Adaptability in Deep Learning training Pipeline for Autonomous Cars

Autonomous driving deep-learning models are trained on a dataset from a particular egocar. We hypothesize the performance will deteriorate when used on a different egocar. In this project, we study suitable imitation learning pipelines for autonomous car that will improve its performance when deployed on different egocars.

People

Data/Codes

Publications

Media


Video segments showing performance of the models deployed on different ego-vehicles.