This Imitator Could Help Seniors With Motor Skills

“Complex diagnostics tell us what is wrong with the brain and muscles.”

An octogenarian reaches for a glass of water on the table. Because of miscommunication between the aging brain and muscles in the hand, and the time-delay in the brain signal reaching the hand—a common occurrence for older adults – he overreaches and knocks the glass to the floor.

Now, a robot manipulator that has been trained and configured in a San Diego State University lab to mimic the human brain may eventually help avoid such mishaps by bringing stability and safety in motion.

Peiman N. Mousavi’s research team has spent four years working on algorithms that can successfully command the robot to mimic the human brain and the jerky actions of an elderly person. This has helped advance the understanding of why the miscommunication and subsequent time-delay happens between the brain and muscles. 

Mousavi, a robotics and control researcher and assistant professor of mechanical engineering at SDSU, has two uncles with Parkinson’s disease. He observed for years how they had difficulty grabbing an object off the table, and began analyzing what was lacking in the loop of actions involved in reaching for objects. 

“Gaps in communication between the brain and muscles in the hands or legs mean that older people will receive the signal much slower, which leads to jerky motions due to ‘actuation delays.’ This can affect their walking and grabbing objects,” said Mousavi.

This delay between command and action is also very common in engineering, so his team’s research has larger implications beyond biomedical uses, for civil and aerospace engineering as well. Time delays can happen when controlling a shuttle orbiting space, with commands traveling long range from earth to space, so when the signal eventually reaches the shuttle, it could potentially cause instability that leads to jerky motions. For buildings with complex networks of water valves, this time-delay technology can help improve the design of smart flow networks. Their research also has implications for U.S military operations. 
Funded by a grant from the National Science Foundation, the study was published July 12 in Automatica, a journal focused on automation and control, and represents the doctoral dissertation work of Mousavi’s student Mostafa Bagheri. 

Miroslav Krstic, senior associate vice chancellor for research and director of a control systems research center at the University of California, San Diego, is the senior author who also guided Bagheri.

The research took four years because the theoretical side of it is cumbersome and involves very complex math and engineering control theory, and translating that to reality and making it compatible with the robot is very challenging, explained Mousavi, principal investigator at SDSU’s Dynamic Systems and Control Laboratory.

“The idea behind this research was to get to complex diagnostics which tell us what is wrong with the brain and muscles, then transfer this knowledge to a microchip… in the form of code, which acts as commands when implanted in a senior’s brain, to send timely signals to their muscles to reliably perform an action,” Mousavi said. 

The chip will significantly improve the motions required to perform the action, but it is a couple more steps down the road from current research. Mousavi’s lab has now completed studying the control and implementation processes required to mimic the human brain, and ongoing research is focused on artificial intelligence (AI) techniques which are underway to train the system, and will eventually lead to the microchip.

Bagheri said he undertook the research because “in our world the performance of a system depends on control, command and feedback, but the reality is that we have delay, which could lead to poor performance or even destabilize the system both in humans as we age, and in engineering systems.” 
He studied time-delays ranging from 0.01 second to several seconds. 

“For space missions, we don’t have the chance to make a mistake, we have to be accurate, but there are delays in teleoperations,” Bagheri said. “So we first worked on understanding whether it was input or output delay that we were dealing with, then we developed an algorithm to help the system – or in this case, to help the brain do the action."

“Our goal is reliable, stable, accurate operations in any system, human or mechanical,” Bagheri said.