Final Report
Abstract:
Robots have long been consigned to the realm of science fiction. From Asimov's Three Laws to Terminator to Star Trek to Lucas' C3PO and R2D2, robots have done everything from threaten to wipe out our entire existence, to help pilot a starship through a dangerous asteroid cluster, to help save the universe from unimaginable evil. However, robots have lately begun to approach the level of technology needed for our imagined forays into the unknown. Within the last ten years or so, robots have even been able to move about freely in an environment that included people. Though it will be a long time before we even begin to approach the marvels we see in the movies, at the rate that the technology is advancing, some of these science fiction writers' visions may begin to come true within the next twenty or thirty years – robots may indeed begin to be seen with more regularity in the workplace, doing the dull menial chores humans are reluctant to do. What will the human reaction be? More importantly, how can their impact and noticeability be reduced so that they are no longer a novelty and something worth paying attention to, but blend seamlessly into the background? This is a question worth asking, and it is the main push behind this study. What makes robots noticeable? How can that be reduced so that robots can interact with people without being a distraction? As robots will begin to become more and more common, it is essential we begin answering these questions now, so that we have ready answers and solutions when it is time to integrate robots into our daily lives.
Introduction:
The motivation behind this study comes from several intriguing observations with Lewis the Photographing Robot. Lewis is about four and a half feet tall, and colored like a fire hydrant, with a similar, if larger, shape. One would think that such an object rolling around a room would be hard, if not impossible, to ignore. However, when Lewis was put in such rooms with humans milling around, a startling observation was noted: after an initial period of fascination, most people tended to ignore the giant rolling red trash can. Time after time this was seen – someone would be intrigued, explore the robot for a while, find out that it did little more than roll around and take pictures, then promptly ignore it. Lewis became a background fixture, nothing more startling than a coffee pot in a kitchen, or vacuum cleaner in a closet. This seems counterintuitive – how can a rolling red cylinder not much shorter than the average person be ignored? Yet ignored he was. Why? This is the question the researchers sought to answer in this study.
This is an important study to accomplish because it is foreseeable that robots will become a greater fixture in the average workplace in the not-so-distant future. Robots lend themselves well to such functions as cleaning (witness the robot vaccuum cleaner Roomba), delivering mail, and other such menial, boring tasks. However, the usefulness of robots will be dulled if they are a constant source of distraction for the humans around them. What must be discovered is what makes robots noticeable or unnoticeable, so that robots in the workplace can be as unassuming and undistracting as possible.
To accomplish this, we asked participants to do a simple inventory task that took them up and down a hallway from room to room. As they walked through the room, they passed by, in turn, a robot and a human. Each participant's run through the study was recorded on videotape, and the times it took them to go between the halls was also recorded. With the videotape, we hoped to observe some interesting reactions a naive user would have to seeing a robot (for perhaps the first time). The recorded times would hopefully provide a strong quantitative measure - if the robot was not noticed, there should be no difference in times between the subject passing the human and the subject passing the robot. In this way, we hoped to discover how noticeable the average robot was to the average human.
Study Outline:
The inventory task ran as follows: To begin with, a participant was led into the start room. There it was explained by the project director that the computer, from a website, would be administering the tasks, and that the tasks would include going from room to room to fetch boxes and count what was inside. Also, it was explained that the whole process would be videotaped, and if they had any questions, those were answered. Then the project director left, and the study began.
The first tasks were to count the boxes in each of the three rooms. This portion was not considered to be part of the experiment, but was used merely as a device to familiarize the participant with the layout of the study. The experiment began formally when the participant left the start room to go to room A and bring back the first box.
After that, at prearranged points during the study, the participant would pass in the hall either a human, a robot, or nothing at all. (Due to technical difficulties, the robot used was not Lewis, but a smaller robot known as Aristo.) Their reaction to each was videotaped, and the time it took for them to pass by the entities in the hall was timed on the website. At the end of the study, the participant was asked to fill out an exit questionnaire to see how well he or she was paying attention to the surroundings. Questions included asking for descriptions of both the robot and the human, and how many times both were passed in the hall. In this manner, we hoped to allow for people that were overall less observant than average.
We chose to time the participants as they passed the various entities because we hoped that there would be a noticeable lag while they passed the robot – pausing to stare at such a mechanical marvel takes time, after all. However, if robots were already close to unremarkable, then there would be no lag; this is where the videotape comes into play. Even if there was no quantitative difference between passing the robot and passing the human, there should almost certainly be a difference in their visible reactions – turning to stare at the robot, perhaps, or eyes widening, or some sort of change in behavior than passing the human or passing nothing at all. On the other hand, we did not choose anything else to measure simply because studies like this one have never been done before, and we literally did not know what to look for. It is hoped that the results of this study will allow for more refined research on the topic of human-robot interactions.
Results:
The results I obtained thus far are not enough to be statistically significant. Only three participants have completed the entire study, and while their results were interesting (one participant paused in the doorway to stare at the robot), they are insufficient for any conclusions to be drawn.
Conclusions:
Unfortunately, my internship ran out before I could gather enough data to be significant enough to draw any conclusions. However, I did train an incoming Washington University graduate student to run this study and gather more data, and I do look forward to completing the study sometime during the upcoming school year. It is a study that is worth doing, and I am glad to have been a part of it.