MUMBAI, India, Nov. 13 -- Intellectual Property India has published a patent application (202621118400 A) filed by Dr. D. Y. Patil Institute Of Technology, Pimpri, Pune on October 03, 2026, for An Ai-Based Adaptive Traffic Signal Management System And Method For Optimizing Signal Timing At Heterogeneous Urban Intersections Using Machine Learning And Passenger Car Unit Analysis.
Inventors include V A Sree Vishnu Dattan; Aditya Narode; Parmeshwar Madan; Ajitesha Singh; and Dr. Ajay Chavadekar.
The application for the patent was published on November 13, 2026, under issue no. 46/2026.
Abstract: An AI-based adaptive traffic signal management system and method for signalized intersections carrying heterogeneous vehicular traffic is disclosed. The system acquires per-category vehicle-count, speed, queue-length and delay data (101), derives twelve engineered temporal, cyclic and lagged features (102), and computes a standardized Passenger Car Unit (PCU) using IRC:SP:41 conversion factors (103). A trained ensemble of machine learning regression models, comprising primarily a Random Forest regressor benchmarked against Gradient Boosting, Support Vector Regression and Linear Regression models, predicts said PCU and vehicular delay at successive short intervals with R² exceeding 0.99 and 0.90 respectively (104). Said predicted PCU/flow-ratio is dynamically substituted, in place of a static historical flow ratio, into the classical Webster's signal-timing formula, so as to compute an optimum cycle length and per-arm green-time split in real time (105), which is applied by an adaptive signal controller (106) to actuate the traffic signal head (107). Emergency-vehicle pre-emption and pedestrian-phase activation modules, and integration with municipal Smart City ICCC/ATMS platforms for black-spot correlation, are further disclosed. Field validation at a representative intersection demonstrated a 28-35% reduction in peak-hour vehicular delay and an improvement of green ratio from 17.6% to 35-45%, without physical road widening.
Disclaimer: Curated by HT Syndication.
