← Back to the feed
AI & softwareResearch findings

DrivingAgent: A Framework for Design and Scheduling in Autonomous Driving Systems

Original published 2026.06.11 00:42 KSTAdded to Blackdeck 2026.11.03 22:47 KST

What has changed?

It separates design and real-time scheduling, optimizing each phase independently.

Key points

  1. DrivingAgent proposes a novel agent framework for automating both design and scheduling in autonomous driving systems.
  2. During design, it interprets system architecture to generate code and validate modules via super-network training.
  3. In real-time scheduling, it dynamically orchestrates modules using a lightweight LLM trained with reinforcement learning.

Connect it to your work

The framework automates module development during design and dynamically orchestrates modules in real time using a lightweight LLM.

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. arxivDrivingAgent: Design and Scheduling Agents for Autonomous Driving Systems