Verfahren und System zur Empfehlung von Folgeaktionen bei der Aufgabenausführung unter Verwendung von Multimodalem Lernen

Details

Veröffentlichungs-Nr.

EP4807616

Anmeldetag

14. März 2025

Veröffentlichung

16. September 2026

Rechtsraum

EP

Abstract

A method and system of recommending follow-up actions in a task execution process, the method comprising capturing language-based features from spoken comments made during task execution, acquiring visual data of the task environment using at least one sensor, processing the acquired visual data to detect and classify equipment-related and environment-related features, generating a structured representation of the task execution by combining the classified visual features with the captured language-based features, applying a trained learning model to analyse the structured representation and determine a recommended follow-up action based on learned correlations between prior task execution patterns and follow-up actions, and outputting the recommended follow-up action. The system comprises at least one sensor configured to acquire visual data, a speech recognition module for processing spoken comments, a data storage module for storing acquired data, and a processor configured to classify detected features, generate a structured representation of the task, apply a trained learning model, and output a recommended follow-up action.

Anmelder

Firma

British Telecommunications Public Limited Company

Land

🇬🇧 Großbritannien