Design of spatial interfaces for engineering assembly within a virtual environment for automotive design

This thesis presents studies on the design of a novel two-handed spatial interface for engineering assembly, informed by a number of qualitative studies using a realistic assembly model within a fully working virtual environment (VE). The results show that the two-handed spatial interface has the potential to reduce task-performance times by more than 25%, over an existing one-handed spatial interface. The VE is the IVPS (Interactive Virtual Prototyping System) at University of Leeds, which supports interactive engineering assembly. The main contribution of this research is to demonstrate an improved understanding of task performance for engineering assembly.

By understanding the assembly task-performance through the evaluation of the existing IVPS using a desktop-based interface, the strengths and weakness of the existing interaction techniques are studied. The results strongly suggest that there is a need to know if more expressive spatial interaction could improve the task-performance for engineering assembly within a VE.

By understanding the assembly task-performance through an evaluation of a one-handed spatial interaction model within the IVPS, a number of problems in spatial selection and positioning have been identified. They are the problems of scale (such as selecting a very small feature from a component), slide (such as manipulating constrained components in an assembly), global precision (such as manipulating the entire scene in which some components are long way from the centre of rotation) and related precision (such as manipulating the selected component related to the other components).

A novel cube-based two-handed spatial interface has therefore been designed to overcome these problems in spatial selection and positioning. It assigns to the nondominant hand tasks such as positioning that can be performed by a sequence of 1DOF sub-tasks. This leaves the dominant-hand to perform the tasks such as 6DOF manipulation of assemblies, selection and attachment. This interface uses a physical cube to provide the user with a spatial frame of reference. The evaluation results show that the cube-based two-handed spatial interface has the potential to reduce the task-performance time by more than 25%, over the existing one-handed spatial interface. A tentative hypothesis is finally generalized and offers opportunities for further research.