English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

The separation assurance task will be extremely challenging for air trafficcontrollers in a complex and high density airspace environment. Deepreinforcement learning (DRL) was used to develop an autonomous separationassurance framework in our previous work where the learned model advised speedmaneuvers. In order to improve the safety of this model in unseen environmentswith uncertainties, in this work we propose a safety module for DRL inautonomous separation assurance applications. The proposed module directlyaddresses both model uncertainty and state uncertainty to improve safety. Oursafety module consists of two sub-modules: (1) the state safety sub-module isbased on the execution-time data augmentation method to introduce statedisturbances in the model input state; (2) the model safety sub-module is aMonte-Carlo dropout extension that learns the posterior distribution of the DRLmodel policy. We demonstrate the effectiveness of the two sub-modules in anopen-source air traffic simulator with challenging environment settings.Through extensive numerical experiments, our results show that the proposedsub-safety modules help the DRL agent significantly improve its safetyperformance in an autonomous separation assurance task.

Safety Enhancement for Deep Reinforcement Learning in Autonomous Separation Assurance