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 Tools annual

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 Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

Many real-world applications can be formulated as multi-agent cooperationproblems, such as network packet routing and coordination of autonomousvehicles. The emergence of deep reinforcement learning (DRL) provides apromising approach for multi-agent cooperation through the interaction of theagents and environments. However, traditional DRL solutions suffer from thehigh dimensions of multiple agents with continuous action space during policysearch. Besides, the dynamicity of agents' policies makes the trainingnon-stationary. To tackle the issues, we propose a hierarchical reinforcementlearning approach with high-level decision-making and low-level individualcontrol for efficient policy search. In particular, the cooperation of multipleagents can be learned in high-level discrete action space efficiently. At thesame time, the low-level individual control can be reduced to single-agentreinforcement learning. In addition to hierarchical reinforcement learning, wepropose an opponent modeling network to model other agents' policies during thelearning process. In contrast to end-to-end DRL approaches, our approachreduces the learning complexity by decomposing the overall task into sub-tasksin a hierarchical way. To evaluate the efficiency of our approach, we conduct areal-world case study in the cooperative lane change scenario. Both simulationand real-world experiments show the superiority of our approach in thecollision rate and convergence speed.

Hierarchical Reinforcement Learning with Opponent Modeling for Distributed Multi-agent Cooperation