Learning to Drive from a World Model

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Goff, Mitchell, Hogan, Greg, Hotz, George, Locmaria, Armand du Parc, Raczy, Kacper, Schäfer, Harald, Shihadeh, Adeeb, Zhang, Weixing, Yousfi, Yassine
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913810104713216
author Goff, Mitchell
Hogan, Greg
Hotz, George
Locmaria, Armand du Parc
Raczy, Kacper
Schäfer, Harald
Shihadeh, Adeeb
Zhang, Weixing
Yousfi, Yassine
author_facet Goff, Mitchell
Hogan, Greg
Hotz, George
Locmaria, Armand du Parc
Raczy, Kacper
Schäfer, Harald
Shihadeh, Adeeb
Zhang, Weixing
Yousfi, Yassine
contents Most self-driving systems rely on hand-coded perception outputs and engineered driving rules. Learning directly from human driving data with an end-to-end method can allow for a training architecture that is simpler and scales well with compute and data. In this work, we propose an end-to-end training architecture that uses real driving data to train a driving policy in an on-policy simulator. We show two different methods of simulation, one with reprojective simulation and one with a learned world model. We show that both methods can be used to train a policy that learns driving behavior without any hand-coded driving rules. We evaluate the performance of these policies in a closed-loop simulation and when deployed in a real-world advanced driver-assistance system.
format Preprint
id arxiv_https___arxiv_org_abs_2504_19077
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Learning to Drive from a World Model
Goff, Mitchell
Hogan, Greg
Hotz, George
Locmaria, Armand du Parc
Raczy, Kacper
Schäfer, Harald
Shihadeh, Adeeb
Zhang, Weixing
Yousfi, Yassine
Computer Vision and Pattern Recognition
Robotics
Most self-driving systems rely on hand-coded perception outputs and engineered driving rules. Learning directly from human driving data with an end-to-end method can allow for a training architecture that is simpler and scales well with compute and data. In this work, we propose an end-to-end training architecture that uses real driving data to train a driving policy in an on-policy simulator. We show two different methods of simulation, one with reprojective simulation and one with a learned world model. We show that both methods can be used to train a policy that learns driving behavior without any hand-coded driving rules. We evaluate the performance of these policies in a closed-loop simulation and when deployed in a real-world advanced driver-assistance system.
title Learning to Drive from a World Model
topic Computer Vision and Pattern Recognition
Robotics
url https://arxiv.org/abs/2504.19077