MUMBAI, India, June 26 -- Intellectual Property India has published a patent application (202611058063 A) filed by Smart Joules Private Limited on May 07, 2026, for Universal Automation System For Hvac Control Using Edge-Deployed Reinforcement Learning And Method Thereof.
Inventors include Saksham Dutta; Akshit Singhal; Venu Madhav Arremsetty; and Arjun Premchand Gupta.
The application for the patent was published on June 19, 2026, under issue no. 25/2026.
Abstract: A universal automation system (1000) and method (400) for HVAC control are disclosed. The system (1000) comprises a resource-constrained computing device (100) hosting (402) a reinforcement learning agent (140). A configuration engine (110) and dynamic factory (130) enable a single executable to control diverse HVAC equipment (200) without code modification. A physics-informed initialization unit (150) ensures safe startup by encoding (414) expert rules into the agent's neural networks, while a temporal credit assignment unit (160) compensates for thermal lag by delaying (418) experience storage. The corresponding method involves receiving (404) configuration data to instantiate control components, initializing the agent (140) with physics-based rules, generating (416) control actions to balance comfort and energy objectives, and ensuring fail-safe operation through a safety framework (180). This approach provides a scalable, adaptive, and efficient solution for HVAC optimization on edge devices. Referring FIG. 1 & FIG. 2
Disclaimer: Curated by HT Syndication.
