Intelligent Traffic Management for Urban Congestion: Application of Machine Learning, Neural Networks, and Fuzzy Logic Systems

Rapid urbanisation in Bangalore has precipitated severe traffic congestion, particularly at critical arterial junctions along the Sarjapur–Outer Ring Road (ORR) corridor. This paper presents a comprehensive intelligent traffic management framework that integrates machine learning (ML), artificial neural networks (ANN), and fuzzy logic systems (FLS) applied to empirical classified traffic count data collected at the Devarabissana Halli (DBH) junction on the Sarjapur ORR. A 24-hour turning movement survey yielding 3,109 total vehicles (3,020 PCU) forms the empirical backbone of this study. The observed morning peak-hour volume of 468 PCU (8:00–9:00) and afternoon peak of 268 PCU (14:00–15:00) reveal a distinct bi-modal distribution that challenges conventional fixed-cycle signal control. The proposed system employs an ANN-based short-term traffic volume predictor, a fuzzy inference system for adaptive signal phasing, and a reinforcement-learning agent for intersection-level co-ordination. Simulation results demonstrate potential reductions of 23–31% in average vehicle delay and a 17% improvement in throughput compared to existing Webster-formula optimised cycles. The framework is designed for scalability across Bangalore's 120+ signalised intersections, contributing toward the city's aspirations under the Smart Cities Mission.