Vision-Based Grasping Robot

Abstract

This thesis presents an in-depth analysis of the composition and secondary development of a vision-based grasping robot. It comprehensively examines the hidden problems encountered during the operation process, along with the underlying causes contributing to these problems. Additionally, this thesis studies various strategies to solve these hidden problems. Building upon the identified shortcomings in existing approaches, this thesis advances the field of research by exploring a semi-automated design methodology that can reduce the likelihood of hidden hazards in robotic systems by exposing them to developers throughout the design phase.

Particularly, this thesis focuses on two fundamental capabilities in post-secondary development: item delivery and door operation. During the execution of these tasks in real-world scenarios, several hidden dangers were exposed, which predominantly stem from human factors in the design and development of the robot. These factors primarily concentrate on the design of the perception subsystem. However, certain latent issues inherent in the design process may not be readily apparent or intuitively discernible to the developers. To effectively tackle the aforementioned challenges, the proposed semi-automatic design methodology segments the robot and its operational environment in incremental steps of one cubic centimetre. This segmentation facilitates a detailed analysis of the data, enabling the identification of potential issues. This concept forms the foundation of the methodology. And the complete process of the semi-automatic design methodology encompasses the following steps: requirements analysis, structural design, environment abstraction, perception subsystem design, hidden disease detection, and optimised design.