A logistics healthcare robot transporting medication, sterile materials and other supplies through a hospital will sooner or later arrive at an elevator. It cannot press the button itself, and its manufacturer does not want it to, because that would require every elevator in every hospital to be adapted. The alternative is simple: ask a passer-by for help.
But how can a robot do that in a clear, unobtrusive way, without sound or light adding confusion to an already busy hospital corridor? That was the central question in the graduation research of Lisa Kirkenier, a Communication and Multimedia Design student at Avans University of Applied Sciences.
“Interaction design and experience design are two very important design principles in my study,” says Lisa Kirkenier. “Interaction design is really about how people deal with a design, how they interact with it. User experience design is about the experience a design gives to a user.”
For the robot, that translates directly into practical questions. “People are going to encounter that robot. How will they interact with such a robot? Can a robot talk? Does it make sound? Does it give light? What does that mean?”
The graduation assignment is a collaboration between robot manufacturer Ambyon and the Media Design research group at HAN University of Applied Sciences. Her research question was: how can visual and non-verbal instructions from a logistics healthcare robot be designed to encourage people in an elevator context to help operate the elevator, without changes to the infrastructure?
Robot arm
Ambyon had already experimented with an extendable robot arm that could press the elevator button itself. But from an interaction design perspective, that approach drew criticism. “It is not a robust solution,” says Kirkenier. “It takes a long time before it actually presses the button. That irritates people. And if people are standing in its way, it still has no way of communicating that it needs to get there.”
Changes to the infrastructure are also not a workable solution. “All those different types of elevators would have to be adapted, which means the robot would no longer be scalable. The robot should basically be ready to use in any hospital.”
Observing
The robot itself has very little contact with people, Kirkenier explains. “It goes from the pharmacy, for example, to the treatment room, where the doctor would take the medication out of the robot.” But the elevator is an exception. “The elevator is a public space in a hospital. That is exactly the moment when it could come into contact with older people, doctors, visitors and probably also children who see it driving by.”
Kirkenier first observed the elevator area at Radboudumc in Nijmegen. “The very first elevator unit at Radboud is also a walk-through area. Behind the elevator unit is the blood collection department, so there are many people who do not need the elevator at all and simply walk through.” She saw “people in wheelchairs, catering staff walking with carts, ambulance personnel with stretchers, doctors with hospital beds.”
Those observations led her to an ethical question. “Is it ethically responsible for the robot to ask just anyone such a question? Ambulance personnel who have just come in with a stretcher do not have time to answer a question like that. They are busy passing on all the information to doctors.”
She therefore spoke with hospital staff, from logistics employees to a doctor in the first cardiac monitoring unit. “I work here with people who do not know whether they are going to make it, because it is a very hectic department. There are lights, sounds, all kinds of things that are actually ten times more important than what that robot wants.”
That led to a clear boundary condition. “It must not attract attention by means of sound, because there is already so much sound in a hospital. But it cannot give a light signal either, because that already happens a lot in hospitals too.”
A voice also turned out not to be an option, because a voice suggests intelligence. “And in fact, that robot is not intelligent at all. It is not a social robot. It brings things from A to B, and it should not come across as if it can answer all kinds of questions, because it cannot.”
From animation to projection
Kirkenier initially designed a small screen on the robot with a short animation explaining the request. That worked, provided people paid attention to it. “That screen does not immediately stand out on such a large robot. People understood the message, as long as they took the time to look at it.”
Her solution became a light projection directly onto the elevator button. “What if the robot projects something directly onto the elevator button, so people immediately see it when they walk past the elevator? You see one large light projection, but the mapping of the request for help is also immediately clear, because the robot directly shows: here, this is the button you need to press for me.”
Surveys and desk research she carried out showed that 76 percent of people are willing to help a robot, and even 86 percent in low-effort situations such as pressing a button. “For that, it is very important that the message comes across immediately, because people do not want to first have to figure out the entire message.”
The main reasons for not helping were: “I do not understand what the robot wants” and “I do not trust the robot.” According to Kirkenier, the latter is also related to authorization. “People simply need to know that they are allowed to help it as a visitor.”
Testing
Testing in an operational hospital proved difficult to organize. Instead, she tested in the Experience Center at Radboudumc, a space set up like a hospital, with treatment rooms, elevators and signage. There she observed medical students and interns to see whether they helped the robot when the request for help was projected.
What surprised her most was how many people simply did not respond. “I did one test where the projection was not very strong yet, with windows all along the side of the building and very bright spotlights directly above the elevator, so the projection disappeared. But the robot was still standing there. So many people who were talking to each other or looking at their phones walked straight toward that robot, somehow took a step to the left or right to go around it, but did not show any sign that something was there.”
Kirkenier ultimately delivered a tested prototype, but emphasizes that the research is not finished. “I could continue for another year with everything I found, because testing always gives you new insights.”
Her research mainly shows that the biggest challenge for healthcare robots at the elevator is not technology, but communication: how do you ask something of a human without sound, without light, without a voice, and still clearly enough to be heard?
