MUMBAI, India, Sept. 28 -- Intellectual Property India has published a patent application (202631112280 A) filed by Brainware University on September 18, 2026, for Edge-Computing Based Adaptive Traffic Signal Control System Using Federated Reinforcement Learning.
Inventors include Amitava Podder; Subrata Paul; and Piyal Roy.
The application for the patent was published on September 25, 2026, under issue no. 39/2026.
Abstract: The present invention relates to a smart-city intelligent transportation system, and more particularly to an edge-computing based adaptive traffic signal control system employing federated reinforcement learning. The system comprises a traffic sensing module (102) at each intersection that acquires real-time traffic density data; an edge computing unit (104) hosting a local reinforcement-learning signal-control agent (106) that computes optimal signal-phase timing based on locally stored control- policy model; a signal control unit (110) that applies the computed timing; an emergency-vehicle detection module (116) that triggers signal pre-emption for approaching emergency vehicles; and a federated learning aggregation server (108) that periodically aggregates model-weight updates, and not raw sensor data, from a plurality of intersections to compute an updated global control-policy model (130) for redistribution to the said intersections. The invention reduces traffic congestion and control-decision latency, preserves the privacy of locally sensed traffic data, and continues local operation during connectivity loss, as compared to conventional fixed-time or centralized cloud-based traffic control systems. Figure 1 is most representative of the invention.
Disclaimer: Curated by HT Syndication.
