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

Ride-sourcing services are now reshaping the way people travel by effectivelyconnecting drivers and passengers through mobile internets. Online matchingbetween idle drivers and waiting passengers is one of the most key componentsin a ride-sourcing system. The average pickup distance or time is an importantmeasurement of system efficiency since it affects both passengers' waiting timeand drivers' utilization rate. It is naturally expected that a more effectivebipartite matching (with smaller average pickup time) can be implemented if theplatform accumulates more idle drivers and waiting passengers in the matchingpool. A specific passenger request can also benefit from a delayed matchingsince he/she may be matched with closer idle drivers after waiting for a fewseconds. Motivated by the potential benefits of delayed matching, this paperestablishes a two-stage framework which incorporates a combinatorialoptimization and multi-agent deep reinforcement learning methods. Themulti-agent reinforcement learning methods are used to dynamically determinethe delayed time for each passenger request (or the time at which each requestenters the matching pool), while the combinatorial optimization conducts anoptimal bipartite matching between idle drivers and waiting passengers in thematching pool. Two reinforcement learning methods, spatio-temporal multi-agentdeep Q learning (ST-M-DQN) and spatio-temporal multi-agent actor-critic(ST-M-A2C) are developed. Through extensive empirical experiments with awell-designed simulator, we show that the proposed framework is able toremarkably improve system performances.

Optimizing Online Matching for Ride-Sourcing Services with Multi-Agent Deep Reinforcement Learning