Testing robotic companions on a simulated mission to Mars

[From IEEE Spectrum’s Automaton blog, where the story includes many additional pictures and a video]

Romibo, companion robot

Testing Robotic Companions on a Simulated Mission to Mars

By Simon Engler
Posted 26 Sep 2013 | 20:02 GMT

Simon Engler (pictured above with robotic dinosaur Pleo) recently spent four months in a Hawaii habitat that simulated a Mars mission. In this guest post, he tells us about the experience and his research on robotic companions for long-duration space missions.

Imagine living for four months in total isolation with six other humans, all locked up in an advanced facility that simulates life on Mars. That was the goal of the Hawaii Space Exploration Analog and Simulation (HI-SEAS) project, and I was one of the lucky individuals to take part in this experiment.

Nearly two years ago, I ran across a call for participants for the HI-SEAS project, which was to take place 8,000 feet up on the side of the Mauna Loa volcano in Hawaii. I didn’t think twice before applying. My background is in astrophysics and mathematics, but I also served as a combat engineer in Afghanistan for the Canadian Army. I happen to have a fascination with robots, too, and I started building them during my down time in the army. One of my robots, called Prairie Dog and designed to scout for improvised explosive devices (IEDs), is now on display at the Military Museums of Calgary, Canada. Since leaving the army, I went back to the University of Calgary and studied robotics full time. However, this love of robotics is mixed with a love of space sciences. I knew that if I could somehow develop robotic platforms for space applications, it would be the perfect career path.

For the HI-SEAS application, each individual was expected to propose their own research project. I suggested a study in robotic companions as a method to reduce stress in an isolated environment. To my great delight, out of 700 applicants, I was chosen as the Crew Engineer. Mars here we go! Well, simulated Mars, to be precise, but this still promised to be a once-in-a-lifetime experience.

Humans on missions to Mars, asteroids, or beyond, will be separated from friends, families, and social stimuli for months or years at a time. They’ll be in a mentally and physically demanding environment, often with minimal personal space. This has the potential to create constant emotional stress, and it’s important to have some sort of healthy way to deal with that stress. It has been well established that domestic pets can provide a great deal of stress relief and create emotional bonds with their owners. Obviously, it’s not very practical for domestic animals to accompany long-term space missions, but robotic companions might be able to help fill that role.

A robotic companion is a robot that is designed to be easy and compelling for humans to form bonds with. Robotic companions are most often animals or animal-like, and examples include Pleo (a robotic baby dinosaur) and Romibo (a shaggy little robot with big eyes). Rather than interacting with humans through text or voice, robotic companions rely on more abstract techniques that would be familiar to any pet owner, including sound, movement, and touch. Essentially, the hope is that a robotic companion would be able to fulfill much of the emotional space that a real pet would, without requiring resources or maintenance. And if strong emotional bonds are formed from the astronaut to the robot, then there is a potential that the interaction will have numerous substantial positive psychological impact when dealing with stress and loneliness.

The HI-SEAS mission seemed like it would be a perfect opportunity to test robotic companions. The six participants (myself included) are isolated from the real world except by email, and email includes a 20 minute communication delay each way to simulate the time it takes for radio signals to travel to somewhere like Mars. This means waiting a total of 40 minutes before you hear back from mission control with a question or problem. This delay was applied to emails and the Internet, too. We were allotted 8 minutes of shower time per week to save on water supplies. When going outside, we have to put on a space suit which isolates us from our environment even further.

After spending three months on the mission, I started to miss interacting with my dog and cats at home. I missed them even though I was in crowded conditions day and night with five of the most amazing people I’ve met in my entire life. The constant stress of the mission and confined conditions slowly eats away at you. After months in isolation, the people one interacts with in the outside world become more important, and anything that keeps you connected to loved ones is valuable. Caring for and worrying about a robotic companion could serve as an outlet for this need. After all, it’s not an environment that humans are really designed for, so it’s important that crews on long duration missions have as many tools as possible for releasing stress.

In order for a robotic companion to be effective, it has to have the right kind of personality. Look and feel are important, but it’s not enough for a human to just “like” the robot. They have to relate and bond to it, and ideally, to fall in love with it over time as one would with an animal companion. But what kind of personality is suitable for a robot companion? Should the robot require constant attention? Will it find the person and interact on its own, or will it wait for the human to approach it? And how will it know how to interact in different ways at different times, depending on what sort of companionship a human needs?

To answer these questions, Dr. Jean Hunter of Cornell University and Aubrey Shick, the creator of a robotic companion named Romibo, are helping me to analyze the effectiveness of two different types of robot companions with programmable personalities in the context of the HI-SEAS mission. Additionally, we are looking to find what specific kind of robotic companion allows people to bond with it most easily.

My project on the HI-SEAS mission was to design and program robotic personalities and implement a preliminary experimental study. The study tested two types of personalities within two different robotic platforms: a demanding personality requiring attention, and a passive personality that would wait for crew members to initiate interaction. The crew members were assigned a robot for a period of time, uploaded with the aggressive and passive personalities. Each crew member was given a survey to fill out after initially interacting with the robot platform. The survey was used to assess the type of experience the crew member had, and to learn about any changes the person would like to see.

Romibo is a big fat fuzz ball that reminded us of one of Star Trek’s tribbles. Originally created by Aubrey Schick at Carnegie Mellon University for therapeutic interaction with autistic children, Romibo has been modified for more sophisticated interactions with HI-SEAS crew. It’s capable of being programmed through its Arduino board with control sets for eye movement, navigation, and to use sensors that allow for various levels of behavior. For example, if it’s left alone for too long, Romibo will fall asleep. Each time it blinks its eyes, the blink will get slower and slower until the eyes no longer open. Afterwards, Romibo will begin to snore. Should someone come near Romibo, the distance sensors and light sensors will detect the presence of a person, and Romibo will open its eyes to see who’s there and begin interacting.

Pleo is a complex robotic toy modeled after a one-week-old Camarasaurus, a species of plant-eating sauropod from the late Jurassic period. Pleo is capable of learning over time, and of simulating the expression of many different moods and basic animal drives such as hunger, fatigue, fear, and love. Pleo is also capable of exploring its environment and learning from it in limited ways. We expected that Pleo would appeal to crew members on the basis of its complex motor skills, vocalizations, and its capacity to explore.

Of course, there are other robotic companions on the market, such as the therapeutic baby seal robot Paro or Sony’s robotic dog Aibo. However, these relatively expensive platforms were outside the budget of this study.

If robotic companions are found to have a positive effect on stress levels and the emotional well being of crew members, we anticipate that they will help to maintain a high level of overall mental health and performance of space crews on long missions. The preliminary results of this study, in particular our comparison of the effects of the robots’ different interactive styles, will assist NASA to design further tests of robot companions in other long analog missions, in short-duration, high-intensity analog tests such as NEEMO missions, or even on the International Space Station.

The idea behind programming each robot with two types of personalities, a passive personality and an aggressive one, was to see if crew members would bond more easily to one personality over another.

Pleo Passive Personality:

Pleo’s passive personality came in the form of a low interaction level and passive calls for attention. Pleo would make a variety of sounds to indicate that it wanted to be petted, that it wanted to play, or that it was hungry. It would respond positively to being touched and scratched by lifting the area being scratched and vocalizing sounds.  If the robot was hungry, it would indicate this by loud vocal calls. A leaf with an RFID chip on it would be held by the crew member and “fed” to Pleo. Afterwards, Pleo would ask for petting by lifting its head, calling out vocally, and wagging its tail. Pleo would also play a game of ‘tug-of –war’ by biting on a RFID stone on a string and pulling, similar to games domestic dogs play with their owners.

Pleo Aggressive Personality:

Pleo was also programmed with a more aggressive personality, where the level of interaction was increased. The crew member was given a number of food items to feed to Pleo that changed its behavior by making the robot mimic being sleepy or hyperactive. Pleo was allowed to walk around on the habitat floor during this time, and was capable of locating activity through its cameras and microphones.

Romibo Passive Personality:

With the Romibo platform, the passive personality had the wheels turned off and a cushion placed under the bottom of the robot. This made the robot platform more receptive to being held or hugged by crew members. The Romibo passive personality was modeled after domestic cats. Crew members were asked to place Romibo next to them while working on their computers, relaxing and watching a movie, or while reading in bed. This robotic companion was programmed with a number of functions to allow it to interact with crew members. For example, Romibo would request a hug by asking for it verbally (Romibo can speak a few words of English), or blink and change its antenna colors randomly. When petted, Romibo would giggle, fall asleep, and snore for two minutes, or hum a quiet tune. The distance and light sensors were programmed to watch for changes created when a person moved in its vicinity or away. If Romibo sensed movement of shadows, it would assume a person was moving around and it would call out to the person, verbalizing ”over here” or ”give me a hug,” in an attempt to encourage crew members to interact with it.

Romibo Aggressive Personality:

Romibo was also programmed with an attention demanding personality. The wheels of the robot were exposed, and Romibo was able to move around on its own. It would call out for attention if it was left alone for more than five minutes. Also, it was able to move around and detect centers of activity through sound and movement. In this manner, Romibo is constantly trying to stay engaged with the crew members. In addition to this, Romibo was programmed to play games, including following a flashlight and a version of tag, where Romibo would drive away from crew member for a short period of time and then stop. When the crew member approached the robot it would be indicated by the distance sensor, and the robot would move away again to make its escape.

 

For the first five weeks after arrival of the robots, crew members were assigned the task of taking care of one of the robotic companions in three-day shifts. During this time, each crew member spent six full days interacting with each platform. After the assignment period ended, crew members were no longer required to spend time with a robotic companion, but the robots were still active and demanding attention from the crew members. We thought that this “free” time is when any emotional bonds formed with the robotic companion would be revealed, since if crew members found the initial interaction rewarding, they’d continue to interact with the robot, and if not, the robot would likely be ignored or even shunned.

At intervals during the mission, the crew held focus group meetings in which they discussed their experiences with the robots and suggest modifications in the robots’ behaviors. The first meeting occurred after each crew member completed one three-day shift of caring for a robot companion. The second was two weeks after the five-week “assigned care” period, immediately before a one-week break from robot companions. It was difficult to tell if people missed the robotic companions or not during this time, but the expectation was that once the robot companion was returned the crew members’ morale would be noticeably improved.

It’s going to take us some time to conduct a complete analysis of the data from the HI-SEAS mission, however there are some interesting observations I made as a part of the mission myself. First, crew members who interact with animals in their normal lives at home tend to react and bond with robotic companions much more easily and quickly. This becomes more evident when you give a robotic companion to a person who’s had little or no exposure to pets in their everyday lives: there seems to be a sense of confusion to what they are supposed to do with the robot.

Second, people interact with the robot in a variety of different ways. Some would passively interact with it while working on their computers. Others would dedicate their full attention to the companion, and experiment by making it do tricks, or dressing the companion up to run on the habitat treadmill for some imaginary exercise. These kinds of interactions create a great deal of laughter and sense of fun in the habitat, and I think at these times the robotic companions definitely achieved their goals.

Finally, it is important that the robotic personality matches the way the user wishes to interact with the robot. Those who worked with the robotic companion by the computer would like the robot to be quiet but attentive, kind of like how a cat will sit with you when you work, or lay on your keyboard for attention. If the robot was more demanding in these situations, the user would simply become annoyed. It will be important for robotic companions to be able to modify their behavior according to the changing needs of the user.

HI-SEAS will be conducting other isolation experiments over the next three years. I expect to be involved in these next missions as the support team while doing a PhD at the University of Hawaii, and I intend to continue the study on robotic companions. The next steps will be dedicated to developing robotic companions with a large variety of personalities and behaviors with the ability to adapt to the needs of their user. With a little luck, these robotic companions may someday be traveling to the stars, and allowing those within the confines of spaceships and planetary habitat to unwind and relax a little.

[ HI-SEAS ]


Comments


Leave a Reply

Your email address will not be published. Required fields are marked *

ISPR Presence News

Search ISPR Presence News:



Archives