Intellectual Property India Publishes Patent Application for A Computer-Implemented Gaze-Regularized Multi-Modal Imitation Learning System And Method For Autonomous Vision-Based Navigation Of An Unmanned Aerial Vehicle Using Synchronized Depth Perception And Predictive Temporal Representation Learning.

MUMBAI, India, July 24 -- Intellectual Property India has published a patent application (202641085875 A) filed by Mujeeb Rahaman T; Ameen Sha C; and Dr. Saidalavi Kalady on July 13, 2026, for A Computer-Implemented Gaze-Regularized Multi-Modal Imitation Learning System And Method For Autonomous Vision-Based Navigation Of An Unmanned Aerial Vehicle Using Synchronized Depth Perception And Predictive Temporal Representation Learning..

Inventors include Mujeeb Rahaman T; Ameen Sha C; and Dr. Saidalavi Kalady.

The application for the patent was published on July 17, 2026, under issue no. 29/2026.

Abstract: The present invention discloses a computer-implemented autonomous navigation system and method for an unmanned aerial vehicle (UAV) employing a synchronized multi-modal imitation learning framework for real-time navigation in structured and unstructured environments. The system comprises an RGB image acquisition module, a depth acquisition module, a controller input acquisition module, an eye-gaze acquisition module, a synchronized acquisition engine, a multi-modal dataset generation module, a navigation model engine and a flight command generation module. The synchronized acquisition engine acquires RGB images, corresponding depth information, operator gaze coordinates and operator flight control commands within a common acquisition cycle to generate synchronized navigation records, thereby substantially reducing temporal mismatch between environmental observations and corresponding control actions. The navigation model engine generates appearance and geometric feature representations from the synchronized RGB images and depth information, integrates the generated feature representations into a unified environmental representation, regulates feature learning using synchronized operator gaze information, performs temporal sequence processing and predictive representation learning during model establishment, and generates continuous flight control commands comprising forward, lateral, vertical and yaw control components for autonomous navigation. The disclosed system improves navigation stability, obstacle perception, synchronization accuracy, training data fidelity and autonomous operation in previously unseen environments while eliminating dependence on explicit map generation, localization and conventional path planning techniques. In one exemplary implementation, synchronized navigation data are generated within an AirSim simulation environment executing on an Unreal Engine platform using RGB image data, depth information, Beam Eye Tracker gaze data and Xbox controller commands. Fig. 1

Disclaimer: Curated by HT Syndication.