Towards the development of a robotic dexterity assessment framework

Abstract

The importance of flexible manufacturing processes and small-scale assembly is growing due to the increased pace of globalisation and greater consumer demands for product customisation. To address this trend, there has been increased interest in collaborative industrial robots that are better suited to work alongside humans in this changing environment. However, adoption of these robots within flexible manufacturing processes has been relatively slow, which can be, as demonstrated in this thesis, attributed to the current uncertainty around robotic dexterity. Dexterity is a key requirement if collaborative industrial robotics is to be useful within flexible manufacturing processes. However, an effective method for defining and measuring robotic dexterity is currently lacking, which has made it difficult for robotic integrators to introduce collaborative robots in these processes due to challenges in mapping their performance to the task assembly requirements.

The core hypothesis that underlies this work is that the adoption of industrial robots within flexible manufacturing processes can be facilitated by the development of a framework which comprehensively assesses robotic dexterity. This framework should consider the range of dexterous requirements within flexible manufacturing processes, and compliment current manufacturing assessment methods in order to maximise its scope and ease-of-use within the area.

To develop such a framework, this work explores and defines robotic dexterity within flexible manufacturing by abstracting influences from both the human and robotic dexterity literature. The demand for robotic dexterity is found to stem from the surrounding environment, and the classification tables within the Boothroyd-Dewhurst (B-D) design-for-assembly method are shown to comprehensively represent the range of possible operations within flexible manufacturing processes. Accordingly, this work demonstrates that the dexterity of a robotic system within flexible manufacturing can be captured by considering the robotic system's ability to perform the operations identified within the B-D classification tables.

From this consideration of dexterity, the B-D tables are used to compose a robotic dexterity assessment framework that measures a robotic system?s dexterity and determines its potential within flexible manufacturing operations. The framework develops robotic performance metrics which supersede supplier specifications and provide a greater insight into the dexterous ability of industrial robot systems. These metrics consider the dexterous requirements identified by the B-D tables, and their links to these tables simplify robotic integration within flexible manufacturing processes.

The capability of the developed framework is demonstrated using three scenarios commonly encountered within small-scale assembly. For each scenario, the framework provides a structured approach for analysing the scenario and evaluating robotic systems. An initial set of performance metrics is used to estimate the performance of different robotic systems, and physical testing is performed to validate the estimated results. In all scenarios, the developed framework provides an accurate estimate of the robot system's probability of success (PS) and task completion time (CT), which combine to characterise the dexterity of the robot system. These values are compared to normative data from B-D tables to determine the viability of each robotic system relative to human operators. Use of the framework demonstrates that a human operator is the most suitable choice in the majority of the considered scenarios, which supports the current dominance of manual labour within flexible manufacturing processes. Furthermore, the framework permits direct robotic system comparisons and helps to quantify the current gap between human and robot dexterity, which is an invaluable tool for robotic integrators and highlights the framework's potential for adoption within flexible manufacturing.