Chips & computingResearch findings
Doe-1: Closed-Loop Autonomous Driving with a Large World Model
Original published 2024.12.13 03:59 KSTAdded to Blackdeck 2026.11.11 22:17 KST
What has changed?
A closed-loop autonomous driving framework using a large world model unifies perception, prediction, and planning in a single end-to-end system.
Key points
- Proposes a closed-loop framework that improves scalability and decision efficiency over existing open-loop methods.
- Demonstrates effective performance in visual question answering and motion planning tasks on nuScenes.
- Introduces a multi-modal transformer that autoregressively generates perception, prediction, and planning tokens in a unified manner.
Connect it to your work
It generates future predictions directly in RGB space and plans actions using position-aware tokens.
What remains unproven
This research note is based on an abstract; check the original for test conditions and application scope.
Originals used for this note
Links open the original publisher in a new tab.