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

Reinforcement Learning (RL) has been widely explored in Traffic SignalControl (TSC) applications, however, still no such system has been deployed inpractice. A key barrier to progress in this area is the reality gap, thediscrepancy that results from differences between simulation models and theirreal-world equivalents. In this paper, we address this challenge by firstpresenting a comprehensive analysis of potential simulation parameters thatcontribute to this reality gap. We then also examine two promising strategiesthat can bridge this gap: Domain Randomization (DR) and Model-AgnosticMeta-Learning (MAML). Both strategies were trained with a traffic simulationmodel of an intersection. In addition, the model was embedded in LemgoRL, aframework that integrates realistic, safety-critical requirements into thecontrol system. Subsequently, we evaluated the performance of the two methodson a separate model of the same intersection that was developed with adifferent traffic simulator. In this way, we mimic the reality gap. Ourexperimental results show that both DR and MAML outperform a state-of-the-artRL algorithm, therefore highlighting their potential to mitigate the realitygap in RLbased TSC systems.

Bridging the Reality Gap of Reinforcement Learning based Traffic Signal Control using Domain Randomization and Meta Learning