Towards realistic haptic rendering of surface textures

Seungmoon Choi, Purdue University

Abstract

This dissertation presents a series of studies performed on the perceived instability of haptically-rendered surface textures. By perceived instability, we refer to any unrealistic sensations that a user perceives from virtual textures rendered with a force-feedback haptic interface. Our long-term goal is to achieve perceptually realistic haptic rendering of surface textures through a better understanding of factors contributing to perceived instability. Towards this goal, we first quantified the level of perceived stability/instability of a widely-used haptic texture rendering system, and discovered the typical types of perceived instability through psychophysical experiments. Many factors that could potentially affect perceived instability were considered, including texture model parameter, collision detection algorithm, texture rendering method, and human exploration mode. We then characterized the proximal stimuli experienced by a user's hand during the exploration of virtual haptic textures. Several physical variables including position, force and acceleration were measured under conditions where the virtual textures were perceived to be stable and unstable. The proximal stimuli responsible for perceived instability were identified by analyzing measured data in the time and frequency domains, and by comparing the data with known human detection thresholds. Finally, we unveiled the sources of the typical types of perceived instability with additional hypothesis-driven studies based on the measured proximal stimuli. The results of these studies show that (1) the parameter space for perceptually stable haptic texture rendering is too small to be useful for most applications; (2) the typical types of perceived instability (buzzing, aliveness, ridge instability) are due to different characteristics of measured proximal stimuli; (3) a haptic texture rendering system can be passive (therefore stable in the control sense) yet still be perceived as unstable. This dissertation is among the first to demonstrate that perceived instability can result from both device-control instability and inadequate virtual-environment dynamics modeling. It is argued that significant enhancements in both areas are necessary in order for haptic texture rendering to be widely applicable to real-world applications. Future work will develop better control and environmental modeling algorithms for realistic haptic texture rendering.

Degree

Ph.D.

Advisors

Tan, Purdue University.

Subject Area

Electrical engineering|Computer science|Psychology|Experiments

COinS