Year: 2026
Type: dissertation
Abstract: Autonomous driving in complex urban environments requires not only accurat e perception and prediction, but also strict adherence to physical constra ints and safety requirements.While recent deep learning approaches and Vi sion-Language-Action (VLA) models have significantly advanced the capabili ties of data-driven systems, they often overlook vehicle dynamics and envi ronmental constraints, which may lead to physically infeasible or unsafe b ehaviors.This thesis investigates constraint-guided multimodal learning f or autonomous driving, with a focus on integrating explicit physical const raints into trajectory prediction and end-to-end control.
Authors: 葛, 茂寧, GE, Maoning
Language: English
Subfield: Automotive Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7162977153