MUMBAI, India, Aug. 10 -- Intellectual Property India has published a patent application (202641090675 A) filed by Rajarajeswari College Of Engineering on July 25, 2026, for A Multi-Agent Artificial Intelligence And Iot-Based Autonomous Framework For Intelligence Decision Making And Self-Optimizing Smart Infrastructure.
Inventors include Dr. B. Lakshma Reddy; Dr. D. Kirubha; Mr. Rajkumar Ramasamy; Dr S Jagannathan; Mr. T. Auntin Jose; Dr. A. Richard William; Mr. Vm. Saravanaperumal; Mr. Abubakkar Sithik M; and Dr. Rajashekhar S A.
The application for the patent was published on July 31, 2026, under issue no. 31/2026.
Abstract: ABSTRACT OF THE INVENTION: The present invention provides a Multi-Agent Artificial Intelligence and IoT-Based Autonomous Framework for Intelligence Decision Making and Self-Optimizing Smart Infrastructure. The framework implements a decentralized, multi-agent architecture where each infrastructure component is managed by an autonomous intelligent agent capable of independent decision-making, local optimization, and collaborative coordination with peer agents. The framework comprises three logical processing layers of increasing complexity, each employing agents that utilize statistical and machine learning techniques to resolve situational needs and update knowledge. A cognitive edge network is formed where IoT devices maintain asset shadows representing physical properties, logical properties, context information, and capability definitions. Agents employ multi-agent reinforcement learning (MARL) algorithms to continuously optimize policies and communicate with connected agents, sharing parameter values and negotiating consensus for coordinated system-wide optimization. The combination of local optimization and shared consensus provides a scalable, fault-tolerant, and efficient control framework. The self-optimization capability enables continuous improvement through a closed-loop control paradigm: sensors collect real-time data, agents process data to make decisions, actuators implement control actions, and results are measured and fed back into the learning system. The framework supports integration with intelligent transportation systems where each intersection is managed by an independent AI agent coordinating across a network of intersections using graph neural networks and MARL. In water distribution networks, agents control pumps and valves to optimize flow and pressure while detecting and responding to pipe failures in real-time. The invention provides advantages over existing approaches including fault tolerance through elimination of single points of failure, scalability without system-wide reconfiguration, real-time responsiveness through edge processing, autonomous optimization through continuous learning, collaborative intelligence through knowledge sharing, and adaptability to changing conditions. The framework represents a significant advancement in smart infrastructure management, enabling truly autonomous, self-optimizing systems that can adapt, learn, and improve performance without human intervention across diverse infrastructure domains including smart cities, intelligent transportation, energy grids, water distribution, and industrial automation.
Disclaimer: Curated by HT Syndication.
