Reproduce “Safety-Aware AoI Scheduling for LEO Satellite-Assisted Autonomous Driving” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Kangkang Sun, Junyi He, Juntong Liu, Xiuzhen Chen, Jianhua Li, Minyi Guo

A multi-agent reinforcement learning framework, SafeScale-MATD3, was developed to ensure safety-critical information freshness for LEO satellite-assisted autonomous platoons by dynamically managing dual-dynamic channels and sub-slot handover events. This framework, developed at Shanghai Jiao Tong University, successfully meets a strict 1% violation budget for collision alerts and reduces collision-alert Age of Information by 35% compared to other approaches.

Key insights

  • The cumulative Age of Information (AoI) penalty from consecutive "ping-pong" handovers grows quadratically, making oscillation suppression more impactful than reducing individual outage durations.
  • A fine-grained, two-timescale AoI model is necessary to accurately capture sub-slot handover events and enable verifiable compliance with strict, heterogeneous safety-critical information freshness requirements.
  • Integrating virtual queues with a drift-plus-penalty framework allows for the online enforcement of time-average safety budgets with provable guarantees, providing a runtime signal for control policies.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.