Milos Zefran wants to make the world a better place for people, and he’s doing it with robots. His interest in helping humanity and robotics started in the early 90s when he was a student in Slovenia working in a lab where he was part of team helping paraplegics walk by controlling their muscles with an electrical stimulator.
“I was looking for fields where I could apply math and make a difference in people’s lives,” said Zefran, a professor at the University of Illinois at Chicago’s Department of Electrical and Computer Engineering. “Robotics definitely fit the bill. So I started working in robotics and have been doing it ever since.”
Today, he is leading a team of Ph.D. students in the Robotics Laboratory as they research how to make robots more helpful and allow them to interact with people – especially aging adults. His team is made up of Ehsan Noohi Bezanjani and Bahareh Abbasi, who are researching physical interaction, Sina Parastegari, who is engaged in handover, and Yao Feng and Andrey Yavolovsky, who are working on safety.
“We want to figure out how humans can work together with robots,” said Zefran. “The particular application that we are interested in is how the robots can help elderly people remain living at home so they don’t have to go to assisted living facility.”
According to the Centers for Disease Control and Prevention more than 1.4 million people are living in nursing homes in the United States. However, research shows that people that live at home are healthier, more satisfied, in better shape, and can keep in touch with the family.
“To remain living at home, these people need some help,” said Zefran. “They are not completely independent, and there is a large demand for home care aides for the elderly – people that come to your home and assist you. There is a severe shortage for this workforce, and we want to see if we can use robots to alleviate these problems.”
There are robots that can talk to people and keep them entertained and there are robots that remind people to take medications and attend appointments. However, these robots are not performing any physical tasks in the home, and that is one of the challenges his team is addressing. Providing support for seniors is something Zefran sees as necessity as a rapidly expanding senior population could benefit from new forms of home-assistance technology.
“What we want is a robot that can really help,” said Zefran. “One of the activities that is critical is being able to cook for yourself. That is a major determinant, and can be the difference between someone staying home or having to move to the assisted living facility.”
“So we would like the robot to help someone cook,” he added. “However, the robot cannot just replace you. It needs to be a collaborator, not just a servant to the human.”
Helping and collaborating are the key objectives of the robot. Aging adults need to be engaged, thinking about what needs to be done, and moving around. If the robot does everything for them, the people will deteriorate quickly and it will defeat the purpose.
“People will be working with the robot very closely, moving things together, passing things to each other,” he said. “Handing over a plate from the robot to the human turns out to be a difficult task for robots, and these are the things we are studying.”
Zefran and his team are researching how a robot can smoothly transfer an object to human. The goal is to not let the object fall, and execute a smooth transition without the person having to pull the object from the robot. The robot should understand to let it go. To accomplish this task, the team uses a variety of research tools, and at this stage their main instrument is a data glove.
“The glove has a variety of pressure sensors and the assistant typically wears the glove,” said Zefran. “The sensors measure the pressure exerted, so we can see how you grasp something and how strong the forces are. One subject wears this and they do various things in the kitchen – take a pot together to the stove, setting table together, and pass forks and knives.”
Zefran is interested in measuring force exchanges with this glove, and the action and nonverbal communication they convey. All of the participants in the experiments use physical interaction and often communicate through that physical contact without saying a word.
“There are many ways to communicate and one way is through physical interaction,” he said. “When we carry something, we constantly push and pull to try to express where we are with a task. We are trying to understand these actions and figure out whether there is some sort of language in this physical forces that are being exchanged.”
“We have humans do many of these things and we are looking at how this sort of interaction with forces works together with language interaction,” said Zefran. “These are two ideas we have to bring together to understand what is going on.”
Apart from physical interaction in collaborative activities, students are researching other aspects of collaboration, safety being the most important. They are using a life-size robot to study how it can safely hand over an object.
“Whenever we develop technology that works close with humans, we are always concerned about safety,” said Zefran. “We are working on how we can test different systems and make sure they cannot hurt humans. We can come up with a mathematical expression that formalizes requirements such as that the robot should never exert a force larger than a specified threshold or that it should never hit a human faster than a certain speed.”
In another part of the lab, a miniature autonomous car captures a realistic example of how a real autonomous car works, without the dangers of everything around it and the difficulty to buy and manage a real car. Instead of testing the ideas on a complicated and expensive system, the little car with the sensors and laptop is a much simpler system.
“Autonomous cars are robots because they need to sense, think and move. Not all robots need to look like a human,” said Zefran. “What if a sensor fails? Will that make the car accelerate instead of stopping when there is a pedestrian in front of it? These are things we are testing and want to prevent with our robots.”
The complex robots Zefran and his team are developing will not be available for several years. They must have a full understanding of the fundamental principles behind each robot task before moving to the next level.
“First we need to understand touch and sense, and then we can build robots that effectively use them,” said Zefran. “Learning how to act is the next level for the robots. We are not doing that right now but are excited about the opportunities for new developments in the future.”
Learn more about Milos Zefran’s research and ECE’s Robotics Lab at Milos Zefran.
By David Staudacher, UIC