MUMBAI, India, Sept. 28 -- Intellectual Property India has published a patent application (202641111042 A) filed by Institute Of Engineering And Management, School Of University Of Engineering And Management on September 16, 2026, for Reinforcement Learning Based Controller Design For A Pinned Multi Agent System Framework Via Hierarchical Meta Strategy Learning.

Inventors include Ranadhir Das; Bapi Biswas; Arijit Ganguly; Soham Bangal; Nitu Saha; and Sayantani Das.

The application for the patent was published on September 25, 2026, under issue no. 39/2026.

Abstract: The present invention relates to a reinforcement learning-based controller for coordinated control of a pinned Multi-Agent System (MAS) using hierarchical meta- strategy learning. The proposed framework enables a plurality of autonomous agents to achieve coordinated behaviour under dynamic operating conditions through a combination of graph-based communication, pinning control, reinforcement learning, and hierarchical meta-strategy adaptation. Each agent obtains local state information and selected neighbouring-agent information through a communication topology represented by a graph. One or more selected agents are pinned to a reference trajectory or supervisory controller to propagate global coordination information throughout the multi-agent network. A hierarchical meta-strategy learning module determines suitable control strategies at a higher decision level, while an agent-level reinforcement learning controller generates low-level control actions. The framework continuously evaluates coordination error, communication conditions, control effort, environmental disturbances, and agent states to adapt the control policy. A reward mechanism jointly considers consensus error, tracking performance, energy consumption, stability-related constraints, and control smoothness. The proposed invention provides adaptive coordination of heterogeneous or homogeneous agents while reducing dependency on manually designed control laws and enabling improved response to disturbances, topology changes, and varying operating conditions.

Disclaimer: Curated by HT Syndication.