Page 23 - ME-07-E

In early March, the Faculty of Mechanical gery to implant electrodes in the brain; and In the first project funded by the Ros-

Engineering held an unveiling ceremony non-invasive interfaces. The most common si Ascoli Foundation, we developed a

for the Brain-Computer Interfaces for Reha- non-invasive interfaces are EEG-based in- brain-computer interface that controls the

bilitation (BCI-Rehab) lab, named after the terfaces, which are noteworthy in terms of position of the hands and automatical-

late Dr. Maria Rossi Ascoli. The lab focuses comfort and use. ly corrects errors by identifying the ErrP.

on developing non-invasive brain-comput- The original aim of brain-computer in- The interface contains two categorizers.

er interfaces based on EEG measurements terfaces was to allow people with spinal The first is a hand movement categorizer,

from electrodes attached to the scalp, and injuries and locked-in syndrome (LIS) to which identifies three commands: chang-

on improving their accuracy. operate a computer or machine directly ing the position of the right hand, that of

Participating in the ceremony were Ben with their brains. The development of EEG- the left hand, or leaving both hands in the

Zion Feigelson, Esq., administrator gener- based interfaces opened up the possibility same position. The second categorizer fo-

al and official receiver at the Ministry cuses on errors: Was there an error,

of Justice; Gali Gross , Esq., deputy and if so, what kind? Was the error

administrator general; Liora Govrin, committed by putting the hand in a

Esq., deputy director of the Estates different position? Or by in changing

and Trusts Department; and Carmit the position of one of the hands, in-

Moses-Cohen , Esq. Present from the stead of leaving the position of both

Technion were Ms. Zahava Laniado, unchanged? Identifying the ErrP and

deputy director general of operations correcting the actions that provoked

at the Technion and trustee of the foun- it improved the performance of the

dation; Prof. Alon Wolf, dean of the interface by around 7%.

Faculty of Mechanical Engineering; re- Thinking and In the second project, we re-
tired Prof. Reuven Katz, one of the lab’s searched the ErrP provoked when
founders and director of the Technion’s there are disturbances in the action

innovation and entrepreneurship pro- Operating of an exoskeleton. Active exoskel-
gram; Prof. Miriam Zacksenhouse, etons are developed to assist dis-
director of the lab; the lab’s research abled and abled-bodied people in

and engineering team: Dr. Miri Ben- performing strenuous physical ac-

yamini, Dr. Igor Demchenko, and M.Sc. The recently opened Maria Rossi tivities. One of the problems is that
students Benjamin-Israel Kaplan and Ascoli Brain-Computer Interfaces for the controller of the exoskeleton
Tamar Shavit; the lab’s engineering Rehabilitation (BCI-Rehab) Laboratory doesn’t always correctly predict the
team: Ariel Bar Yehuda and Kfir Cohen; desired movement of the user, and

and guests. works to reinforce the accuracy of this can hinder rather than help. The
The lab has received a donation interfaces such as these and integrate goal of this project was to improve
the control of active exoskeletons
from the estate of the late Dr. Maria them into systems of exoskeletons so that they help rather than hinder
Rossi Ascoli for the past five years. Dr. and robots that work with people the user’s activity by identifying the
Ascoli, a Jewish-Italian psychiatrist, left

an estate in the hands of the Deputy ErrP provoked in the second case.

Administrator General of Israel she re- By training the convolutional neural

quested be directed into a foundation networks (CNN), we can successfully

for research fellowships and projects in the of using brain-computer interfaces for re- identify haptic potential error with an accu-

medical, science and social fields – includ- habilitating patients – after a stroke, for ex- racy of 75%.

ing fields related to the physical and men- ample – and even for empowering healthy We rare continuing to work on these two

tal well-being of older people and people people. Possible uses include wheelchairs, projects: First, we are acting to increase the

with disabilities. The estate’s contribution robots, drones and games. number of commands that the brain-com-

to the Brain-Computer Interfaces for Re- The main obstacle preventing the broad puter interface can identify and its accu-

habilitation Laboratory has enabled us to use of brain-computer interfaces is the racy; second, we are acting to incorporate

upgrade the lab and advance the research high error percentage (20%-30%). One of the ErrP in learning with reinforcements to

it conducts. the most promising methods of reducing improve the control of exoskeletons and

the error percentage is based on identify- robots that work with people. Outstanding

Brain–computer interfaces create a di- ing brain activity triggered when the user students who are interested in joining this

rect channel of communication between notices that the action being performed is challenging research (especially those with

the brain and the outside world, with no incorrect. The EEG activity that is triggered a background in signal processing, ma-

need for physical movement. Interfaces in response to errors is called “error-relat- chine learning, control, or EEG experimen-

like these are based on two main technolo- ed potential” (ErrP) and can be identified tation) are invited to contact Prof. Miriam

gies: invasive interfaces, which require sur- and used for correcting errors. Zacksenhouse (mermz@technion.ac.il). 

Faculty of Mechanical Engineering | MEgazine | 23