Deep Reinforcement Learning for Sustainable Urban Mobility: A Bibliometric and Empirical Review

Area of Science:

• Computational Integration Framework (CIF)
• Artificial Intelligence (AI)
• Smart City Applications
• Urban Mobility

Background:

• Bibliometric analysis identifies urban mobility as a mature domain for AI implementation.
• Existing AI frameworks lack systematic integration for smart city applications.

Purpose of the Study:

• To provide an empirical basis for the Computational Integration Framework (CIF).
• To demonstrate the feasibility of AI, specifically Deep Reinforcement Learning (DRL), in optimizing urban mobility.
• To bridge AI techniques with urban application domains.

Main Methods:

• Large-scale bibliometric analysis for domain selection.
• Development and validation of a Computational Integration Framework (CIF).
• Implementation of a Deep Reinforcement Learning (DRL)-driven traffic signal control system.

Main Results:

• CIF validates DRL for sustainable urban mobility, achieving ~48% reduction in average wait time and >30% increase in traffic efficiency.
• Federated DRL maintains 96% of central performance while preserving data privacy.
• Significant reductions in fuel consumption and CO2 emissions were observed.

Conclusions:

• Evidence-based domain selection via bibliometrics is effective.
• CIF serves as a viable AI decision support bridge for urban applications.
• DRL is computationally feasible for sustainable urban mobility, aligning with global sustainability goals.