Goal-Oriented Autonomous Driving with Large Language Models
Bringing LLMs into an end-to-end driving pipeline to give it a goal and focus its attention on what matters in the scene.
An end-to-end, goal-oriented approach to autonomous driving using large language models.
We integrated LLMs into the transformer-based feature extraction and motion estimation steps of an end-to-end autonomous driving pipeline (UniAD). The language model provides the driving goal and guides the network’s attention toward the important objects in the scene.