E-Thesis 46 views 30 downloads
Development and Implementation of Intuitive Human-Machine Interaction System / SHUO ZHU
Swansea University Author: SHUO ZHU
-
PDF | E-Thesis – open access
Copyright: the author, Shuo Zhu, 2025. Distributed under the terms of a Creative Commons Attribution Non Commercial 4.0 License (CC BY-NC 4.0).
Download (14.3MB)
DOI (Published version): 10.23889/SUThesis.71084
Abstract
Robotics has been well-developed in various fields in recent years. With the widespread application of intelligent machines, the interaction between humans and them is particularly important. Simple and reliable interactive systems are the general trend of future development. This thesis focuses on t...
| Published: |
Swansea
2025
|
|---|---|
| Institution: | Swansea University |
| Degree level: | Doctoral |
| Degree name: | Ph.D |
| Supervisor: | Li, C., and Giannetti, C. |
| URI: | https://cronfa.swan.ac.uk/Record/cronfa71084 |
| Abstract: |
Robotics has been well-developed in various fields in recent years. With the widespread application of intelligent machines, the interaction between humans and them is particularly important. Simple and reliable interactive systems are the general trend of future development. This thesis focuses on the development and implementation of serious intuitive human-machine interaction systems and explores the application of technology in graph classification, robot learning, droplet detection and robot control. First of all, the real-time observation of the intelligent machines to the outside world is particularly important. A real-time double emulsion droplet detection system based on colour space segmentation and Hough transform is proposed, which can accurately detect the contour, size and generation frequency of droplets, and provide support for quality monitoring in industrial production. In terms of graph classification, the research introduces a target-unbiased meta-learning algorithm to solve the graph structure classification problem under small sample data, demonstrates its excellent generalization ability, and improves the efficiency of image processing and recognition in human-machine interaction systems. In robot learning, the model optimized by extreme learning machine(ELM) combined with beetle antennae search(BAS)is studied, which significantly improves the accuracy and learning speed of trajectory pre-diction of KUKA iiwa robot and enhances the operational flexibility and real-time performance of the robot. The research also proposed robot path planning based on a fault-tolerant motion planning algorithm to achieve automatic drawing tasks. These studies demonstrate the important role of machine learning algorithms in robot control and planning, and significantly improve the robot’s ability to interact with the environment. In addition, the research also developed a flexible sensor data glove that can be used for remote control, further improving the naturalness and real-time nature of human-machine interaction. By integrating manual algorithms with machine learning technology, the thesis provides innovative solutions for the development of intelligent human-machine interaction systems and promotes technological progress in this field. |
|---|---|
| Keywords: |
Robotics, artificial intelligence, interactive systems, neural networks, path planning |
| College: |
Faculty of Science and Engineering |