MUMBAI, India, Aug. 31 -- Intellectual Property India has published a patent application (202641099023 A) filed by St. Peters Engineering College on August 15, 2026, for A Multi-Agent Reinforcement Learning System For Cooperative Autonomous Decision Optimization.
Inventors include Mr. Redapangu Saidulu, Assistant Professor In; Dr. Gurunadham R, Associate Professor In Department Of Csedata Science, Malla Reddy College Of Engineering And; Mrs. Lingathoti Eswaramma, Assistant Professor In; Mrs. Beeram Bhuvaneswari, Assistant Professor In; and Dr. K. Srinivasulu, Associate Professor In Department Of Ece, Ellenki College Of Engineering And Technology, Patelguda, Near Bhel, Hyderabad, Telangana, India -.
The application for the patent was published on August 28, 2026, under issue no. 35/2026.
Abstract: The present invention relates to a Multi-Agent Reinforcement Learning System for Cooperative Autonomous Decision Optimization, designed to enable multiple autonomous agents to learn, communicate, coordinate, and execute optimized decisions in dynamic, uncertain, and partially observable environments. The system comprises an environmental data acquisition module, agent state processing module, adaptive inter-agent communication module, cooperative reinforcement learning module, reward evaluation module, policy optimization module, decision coordination module, and autonomous action execution module. The environmental data acquisition module acquires real-time information from sensors, connected devices, databases, communication networks, and other data sources. The agent state processing module converts the acquired information into representative states for individual autonomous agents. The adaptive communication module selectively exchanges relevant state, action, task, and environmental information among agents to facilitate efficient coordination while reducing communication overhead. The cooperative reinforcement learning module enables agents to learn optimized policies through continuous interaction with the environment and other agents. A cooperative reward mechanism evaluates both individual agent performance and collective system objectives, thereby encouraging decisions that improve overall system performance. The policy optimization module dynamically updates agent policies based on environmental feedback, accumulated experiences, and interaction outcomes. The decision coordination module identifies potential conflicts, resource competition, task duplication, and incompatible actions and determines coordinated actions before execution. The system establishes a continuous perception, communication, learning, decision, execution, and feedback cycle, enabling adaptive and scalable autonomous operation. The invention provides improved cooperative decision-making, resource utilization, adaptability, convergence, and operational efficiency for applications including autonomous vehicles, robotics, logistics, intelligent transportation, smart manufacturing, and distributed autonomous systems.
Disclaimer: Curated by HT Syndication.
