← Back to the feed
AI & softwareResearch findings

GenAD: A Generative Framework for End-to-End Autonomous Driving from Raw Sensor Input

Original published 2024.02.18 17:21 KSTAdded to Blackdeck 2026.11.14 22:43 KST

What has changed?

GenAD proposes a generative framework that directly predicts future evolution of ego and surrounding vehicles from raw sensor inputs.

Key points

  1. GenAD proposes a generative framework that directly predicts future evolution of ego and surrounding vehicles from raw sensor inputs.
  2. It transforms surrounding scenes into map-aware instance tokens and learns trajectory distributions in a structural latent space to model ev
  3. Extensive experiments on nuScenes show state-of-the-art performance in vision-centric end-to-end autonomous driving with high efficiency.

Connect it to your work

It transforms surrounding scenes into map-aware instance tokens and learns trajectory distributions in a structural latent space to model traffic evolution.

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.

  1. arxivGenAD: Generative End-to-End Autonomous Driving