A Mobile, Map-Based Tasking Interface for Human Robot Interaction

Abstract

Effective communication and user input are crucial to mobile human-robot interaction tasks, especially in cases where robots are deployed to assist with disaster relief, emergency response, or military applications. This thesis details the design and implementation of a mobile, map-based tasking interface that leverages novel communicative and interactive methods to facilitate the completion of mobile Human-Robot Interaction tasks. In particular this interface employs Multi-Touch capabilities on an intermediate sized device, such as a tablet computer, to permit individuals to task and supervise robots. The primary objective is to develop an interface that will allow a mobile supervisor to oversee a team of robots while interacting with other humans. This thesis focuses on the development and evaluation of two fundamental components necessary to achieve the primary objective.

A Communications Mode was developed to facilitate the rapid and effective communication of relevant information between individuals. A user evaluation was conducted to determine the usability of the Communications Mode compared to a paper-based and an audio-based communication method. Findings from this user evaluation determined that the Communications Mode was superior to the audio-based method and was at least as good as the paper-based method.

This thesis also explores the fundamental differences between mouse-based and touch-based interaction when performing drawing tasks by comparing error rates and drawing speed using each of the two interaction methods. A user evaluation determined that, for drawing tasks, users commit more errors using touch-based interaction, but also complete tasks faster than when using mouse-based interaction due to a higher average drawing speed.

Finally, this thesis proposes and evaluates four methods for the reduction of user input, either via mouse or Multi-Touch, during drawing tasks. Data reduction is essential because users frequently supply unnecessary information, in the form of excess points on the screen, while performing a drawing task. Reduction methods were compared by the resulting error generated compared to a user’s true input and the amount of reduction each method afforded. A user study determined that, of the four methods, Cubic Iterative reduction yielded the best reduction with an acceptable rate of error.