MUMBAI, India, Aug. 10 -- Intellectual Property India has published a patent application (202641090993 A) filed by Vellore Institute Of Technology on July 27, 2026, for System And Method For Explainable Precision Agriculture Using Causal Deep Reinforcement Learning And Generative Counterfactual Recourse.
Inventors include Vetriveeran Rajamani; and Ramitha S.
The application for the patent was published on July 31, 2026, under issue no. 31/2026.
Abstract: ABSTRACT SYSTEM AND METHOD FOR EXPLAINABLE PRECISION AGRICULTURE USING CAUSAL DEEP REINFORCEMENT LEARNING AND GENERATIVE COUNTERFACTUAL RECOURSE A precision agriculture system (100) for causal deep reinforcement learning and generative counterfactual recourse comprises a data input module (110) configured to receive farm data including NPK levels, soil moisture, and weather predictions from IoT sensors and to generate an input hash using a hash-locking mechanism. A structural causal modeling module (120) transforms the hash-locked farm data into a structural causal model identifying causal drivers of plant stress. A strategic deep reinforcement learning agent (130) optimizes for seasonal yield and profit based on an objective reward function. A tactical control layer (140) monitors real-time environmental signals, detects vascular bottlenecks where environmental stressors disrupt causal relationships between resource application and plant uptake, and performs policy override when bottlenecks are detected. A counterfactual risk quantification engine (150) runs twin- simulation of action and inaction scenarios to calculate a counterfactual risk value representing financial cost of inaction. A generative explainable artificial intelligence module (160) transforms structural causal modeling output into a natural language implementation blueprint comprising actionable steps.
Disclaimer: Curated by HT Syndication.
