Enhancing Human-Robot Collaboration Through a Multi-Module Interaction Framework With Sensor Fusion: Object Recognition, Verbal Communication, User(s) of Interest Detection, Gesture and Gaze Recognition
Loading...
Authors
Paul, Shuvo Kumar
Issue Date
2024
Type
Dissertation
Language
Keywords
Active Speaker Detection , Computational Interaction , Human Robot Interaction , Interaction Modality , Sensor Fusion , Verbal Instruction
Alternative Title
Abstract
With the increasing presence of robots in our daily lives, it is crucial to design interaction interfaces that are natural, easy to use, and meaningful for a wide range of robotic tasks. This is important not only to enhance the user experience but also to increase task reliability by providing supplementary, task-specific contextual information if needed. Motivated by these goals, we propose a multi-modal framework consisting of multiple independent modules. These modules take advantage of multiple sensors (e.g. image, sound, depth) and can be used separately or in combination for effective human-robot collaborative interaction. We identified and implemented four key components of an effective human-robot collaborative setting, which include: (1) determining Object(s) of Interest location and pose, (2) extracting intricate information from verbal instructions, (3) resolving User(s) of Interest (UOI), and (4) providing gesture recognition and gaze estimation to facilitate natural and intuitive interactions. The system uses a feature-detector-descriptor approach for object recognition and a homography-based technique for planar pose estimation and a deep multi-task learning model to extract intricate task parameters from verbal communication. The User(s) of Interest (UOI) is detected by estimating facing state and active speakers. The framework also includes gesture detection and gaze estimation modules, which are combined with verbal instruction components to form structured commands for robotic entities. Experiments were conducted to assess the performance of these interaction interfaces, and the results demonstrated the effectiveness of the proposed approach.
Description
Citation
Publisher
License
Creative Commons Attribution 4.0 International