MUMBAI, India, Feb. 27 -- Intellectual Property India has published a patent application (202541134202 A) filed by Dhaanish Ahmed College Of Engineering, Chennai, Tamil Nadu, on Dec. 31, 2025, for 'reinforcement learning in autonomous system.'
Inventor(s) include Dr. P. Paramasivan; Dr. K. Senthamil Selvan; Mrs. V. Maria Christy; Mrs. R. L. Shyja; Mrs. S. Jagatheeshwari; Dr. E. Kowsalya; Ms. G. Sahana; Mrs. R. Jesfer; Mrs. M. R. Anitha; and Dr. B. Vaidianathan.
The application for the patent was published on Feb. 27, under issue no. 09/2026.
According to the abstract released by the Intellectual Property India: "This invention discloses a system and method for safely and effectively integrating Reinforcement Learning (RL) into autonomous physical systems. The architecture comprises an RL agent, which includes a policy network for decisionmaking, deployed on an autonomous platform with sensors and actuators. A key innovation is the inclusion of a parallel safety supervisor module that acts as an "action shield," intercepting all actions proposed by the RL agent. This supervisor validates actions against hard-coded safety constraints, preventing the system from executing unsafe maneuvers during its learning and operational phases. The system leverages a high-fidelity simulation environment with domain randomization for safe and efficient pre-training. Furthermore, it supports continuous online learning by storing real-world experiences in a replay buffer for offline policy updates, enabling adaptation to real-world conditions. This combination of shielded real-world deployment and sim-to-real transfer creates a robust framework for developing autonomous systems that can learn complex behaviors, generalize to novel situations, and guaranteed operate within a safe envelope, making RL practical for real-world applications like self-driving cars and autonomous drones."
Disclaimer: Curated by HT Syndication.
