November 4, 2019

It’s no surprise that most of us want to keep a level of self-sufficiency and independence as long as possible as we enter old age. IEEE Impact Creator and Senior Member Ayanna Howard is on a mission to improve the quality of life in dementia patients by encouraging daily movement that improves cognitive retention with the help of machine-learning humanoid therapy robots.

“Humanoid therapy allows us to bring in robots that have the same type of form factor as a human therapist, and fill in the gaps from traditional, in-person therapy sessions,” says Howard.

Utilizing this type of robot will help patients who might not be able to afford daily therapeutic care from a human specialist on a regular basis.

“The human might not be available 24 hours a day, seven days a week — but the humanoid robot can be,” says Howard.

The Dancing Humanoid Robot Increases Cognitive Functioning

Howard and her team are currently developing a humanoid robot focused on helping elderly dementia patients retain their short-term memory for longer increments.

“While there is no cure for dementia, recent studies have suggested that exercise may have a positive effect on the cognitive function of dementia patients,” states Howard’s published research study, co-written by Mariah Schrum and Chung Hyuk Park. “We propose that a humanoid therapy robot is an effective tool for encouraging exercise in dementia patients.”

The interactive humanoid robot, nicknamed Pepper, does a sequence of arm movements and upper body dance moves for the patient to follow along. This type of interaction helps a patient physically and mentally exercise with the goal of improving his or her short-term memory.

“If you can maintain short-term memory and improve it, that means you have a higher possibility of being independent,” states Howard.

Howard believes this type of robot will be best utilized in an assisted living home so more patients will have access to care and reduce facility costs because of a reduced need for human therapy interaction.

See also Humanoid Robots Teach Coping Skills to Children With Autism

The Machine-Learning Technologies Used for Humanoid Therapy

“Humanoid robots not only have to understand our emotional state, but also emote in a responsive way,” says Howard. “It is shown that when we have a robot that is socially interactive in an appropriate way, it increases the amount of time a person interacts with that program.”

Pepper uses machine learning technologies and advanced applications to socially adapt and interact with its patients, which aid in increasing the likelihood of improved cognitive functioning.

As a result, “a robot has to identify what their strengths are and where they need improvement,” says Howard. “As the patient improves, the robot should adapt as well.”

These adaptations are made from actual human observation and then the engineers code those insights into the humanoid robot.

“There are studies that show there is a learning curve for most things,” says Howard. “If we can figure out where you are on the learning curve, we can predict how you’re supposed to learn. If you do something and it doesn’t predict where you should be, then the robot has to adapt what it is doing.”

Facial recognition is another way Pepper is able to assess and interact with a human successfully. Yet, Howard has found in her research that facial recognition does not work as well on children and older adults because there is less data on these demographics. She and her team are currently working on programming the APIs to fit their needs.

“It’s collecting images, codifying it, and then training a typical neural network on that,” says Howard. “It’s a hybrid, so we use the APIs that are already out there. Then that data is imported into our own algorithms that have done more specialized learning on that targeted demographic.”

Howard hopes to publish an article further outlining this research on elder facial recognition by the end of the year. Pepper is still currently in the research stage, and Howard predicts the work will not be done until next year.

Howard’s work in humanoid therapy robotics looks promising for older adults who suffer from dementia. By dancing with Pepper, Howard and her team hopes to see an increase of cognitive retention and independence in patients living in assisted living homes as a result.