Abstract

The usefulness of vibrotactile feedback as a channel to communicate information has been widely recognized. However, most of the recent work on this subject uses actuators that are either too expensive or too large for many practical applications. This thesis explores the generation of vibrotacatile feedback patterns using a simple, inexpensive eccentric-mass motor that is constrained to use a constant voltage and a low on/off switching frequency. In particular, it explores the pulse timing method, which utilizes the slow transient response of the eccentric-mass motor to calculate pulse and rest times for an arbitrary envelope waveform shape. Several hardware tests were performed to (1) obtain a model of the transient response and (2) to verify that the resulting vibrations match the patterns predicted by the pulse timing method. Two custom built devices consisting of an eccentric-mass motor and a rigid housing were used in addition to a Wii remote. Vibrations for each device were measured and compared to the pattern predicted by the pulse timing method when the device was sitting on a table top and when the device was held in the hand. Results indicate that the vibrations match the predicted patterns very well in both cases. It was also determined that error in the motor's transient response model will result in some error between the measured and predicted vibrations. To assess whether this error affects perception of the intended envelope waveform, a study was performed in which users were asked to identify the envelope waveform of vibration patterns created using curve-fit models that contained various levels of error. An analysis of variance revealed that error in the curve-fit will have an effect on the perception of the envelope waveform if the error is large. Two more user studies were performed to determine the perceptual space of patterns generated using the pulse timing method, and to determine whether users could identify the meanings encoded within vibration features. The Perceptual Space study used a cluster-sorted Multi-Dimensional Scaling analysis to determine that envelope waveform, roughness (deviation from the envelope waveform), and amplitude are vibration features that may be used to encode information. Using these features, participants were presented with vibrations that contained GPS navigation instructions similar to those used in a car, and were asked to identify the associated meaning. Users were able to correctly identify all three features with an average accuracy of 80.6%, and were able to correctly identify the envelope waveform and roughness with accuracies of 96.9% and 94.5% respectively. These results are evidence that the pulse timing method (and eccentric-mass motors in general) are capable of generating complex vibrotactile feedback patterns that can be uniquely identified.

Degree

College and Department

Ira A. Fulton College of Engineering and Technology; Mechanical Engineering

Rights

http://lib.byu.edu/about/copyright/