Bionic arms controlled wirelessly by people’s thoughts are coming closer to reality as the result of research at the University of Minnesota that seeks to eliminate the need for risky surgical brain implants in order to work.

Researcher Bin He and colleagues reported on Wednesday the successful use of sensors in a cap worn on the head that interpret brain signals and instruct a robotic arm to make corresponding movements.

Tests with eight study subjects found their thoughts could instruct an arm to complete a variety of real-world tasks, such as picking up a block and putting it on a rack.

“This is the first time in the world that people can operate a robotic arm to reach and grasp objects in a complex 3-D environment using only their thoughts without a brain implant,” said He, who directs the U’s Institute for Engineering in Medicine and its Center for Neuroengineering.

“Just by imagining moving their arms, they were able to move the robotic arm.”

While other clinical studies have successfully used a “brain-computer interface” to control a robotic arm, they required the use of implants that would be close enough to the neurons — the thinking cells of the brain — to record their signals. Attempts to use external sensors were unsuccessful because the skull created too much interference for the sensors to pick up the neural signals.

He’s innovation was the development of software that allowed the sensors to ­filter the interference and ­correctly interpret the movement instructions the brain was ­trying to send.

“We figured out how to pick up the real brain signal out of the huge background noise ... and decode its intention,” He said.

Robotic arms already are options for amputees whose nerves can carry signals from the brain as far their joints. Surgeons then attach the nerve endings to the prosthetic devices and teach the users how to make their robotic arms do what they want. A 13-year-old from Westbrook, Minn., was the youngest patient to ever receive such a prosthetic system from Advanced Arm Dynamics in Maple Grove last fall.

But that technology doesn’t work for people whose nerves have been damaged to the point that brain signals don’t reach their extremities — people with severe spinal cord injuries or strokes. Hence the need for a device linking a computer directly to the brain — either with or without a brain implant.

The trouble with implants is the increased costs and risks associated with brain surgery, He said. “If you want to benefit a huge amount of people in this society, a noninvasive option without surgical implant would be much, much more attractive.”

To test the experimental technology, subjects at the U wore the sensor caps, which were connected via a wire to a robotic arm on a table. (He soon wants to make the process wireless.) They succeeded in 80 percent of attempts to pick up objects from fixed locations and in 70 percent of attempts to place the objects on a shelf.

The human tests followed a variety of basic science discoveries, as well as other clinical trials by which brain sensor technology operated a wheelchair or caused a quadcopter to fly.

He said he was among the first to investigate whether brain signals were the same for people when they actually moved their limbs vs. when they just imagined moving their limbs. The identification of key differences hastened the development of a system that could interpret the brain’s motion signals, even in the absence of a limb.

Subtle neurological differences exist in the way people think and instruct their limbs to move, so the robotic arm software is modified for each person to account for those individual differences.

The U test results were published in the latest edition of Scientific Reports, a subsidiary of the journal Nature.