Researchers have harnessed the power of thought to guide a remote-control helicopter through an obstacle course.
The demonstration joins a growing number of attempts to translate the electrical patterns of thoughts into motions in the virtual and real world.
Applications range from assisting those with neurodegenerative disorders to novel modes of video game play.
The research in the Journal of Neural Engineering uses a non-invasive "cap" to capture brain electrical activity.
It is not the "mind-reading" of fiction. The approach, and others like it, require that an electronic system be "trained" to recognise patterns in an electroencephalograph - a map of electrical activity.
Those thoughts, such as that of making a fist with the left hand, are then correlated with motions of the helicopter - in this case to the left.
The electroencephalograph remains a chaotic and largely indecipherable mess of electrical signals, but those related to motion - or the mere thought of it - have proven to be comparatively strong and repeatable.
When researchers can access the brain directly - with probes or implants - they can focus on more precise areas of brain activity, at the source.
Bin He, director of the University of Minnesota's Institute for Engineering in Medicine and senior author on the new research, and his team have been working toward the helicopter experiments for some time, writing in Plos One in 2011 of similar trials using a virtual helicopter.
For the current work, five participants were selected to wear a simple "cap" that held 64 electrodes, using it to "teach" the computer the brain patterns corresponding to thoughts of movement - clenching of the left and right fist for turning left and right, clenching both fists to go up, and doing nothing to go down.
Then the computer was set up to run the helicopter over wi-fi, with only the participant's thoughts at the controls.
The copter was made to reliably fly through an obstacle course in the university's gymnasium - participants' success rates were as high as 90% in obstacle avoidance.
The new study breaks ground in the autonomous control of a free-moving robot, one of a number of efforts in this direction.
And Prof He said he was convinced that the "non-invasive" approach to gathering the power of thoughts has wider long-term appeal.
"The ultimate application really is to benefit disabled patients who cannot move or patients that suffer with movement disorders," Prof He said.
"We want to to control a wheelchair, and turn on the TV, and most importantly - this is my personal dream - to develop a technology to use the subject's intention to control an artificial limb in that way, and make it as natural as possible," he told BBC News.
Even control of a household robot could become possible, and Prof He thinks that applications should not be limited to those with movement limitations.
"The brain-computer interface technology may ultimately not only help disabled patients but may also help the healthy population... not to restore loss of function but to enhance function beyond what we can accomplish."