Until recently, direct brain-computer communication was a topic known only from science fiction novels and movies. Along with the development of brain imaging technology, the apparatus enabling the recording of brain waves has become many times cheaper and miniaturized, and thus more accessible. We decided to check whether it is possible to prepare software that correctly recognizes the architect’s mental commands and interprets them as program functions.
In the following project, we use a device Emotiv-Epocx, a 16-channel electroencephalograph with a measurement frequency of 256 samples per second. The operation of the device is quite simple. Placed on the head, it measures the electrical voltage at 16 points on the scalp. By comparing signals in different parts of the head, it is able to read brain waves. The device is wireless, lightweight, communicates via Bluetooth and has a battery that allows for continuous operation up to 9 hours.
Our goal was to control BIM (Autodesk Revit) software using an EEG device. But, how is it possible to achieve it? By breaking down the problem into the components, it can be described as follows. The signal sent by the device is cleaned and filtered in real time and saved as 16 graphs of brain waves. Our condition, behavior or activities influence its form. When performing repeated activities, the graph also looks similar, which makes it possible to identify a given human activity.
The person with whom we tested our solution repeated on the monitor screen such activities as increasing the number of stories, decreasing, leaving the number of stories unchanged. During the operation, cerebral activity was monitored and appropriately assigned to functions. Then, the “trained program” (machine learning model, to be exact), was able to recognize, based on the brain waves, what activity the architect would like to perform.
Novel communication method
After training the program can recognize 3 architect’s brain activities, a software then calls the given function inside Revit, reading the number of storeys and depending on the incoming command, it updates the height of the building by adding, subtracting or leaving the number of storeys unchanged. The designed system worked with high efficiency, allowing up to a dozen commands to be executed consecutively without errors, while the software response time was less than 1 second.
The designed system proves that it is possible to control architectural design software with a brain waves. The architectural design process itself is highly complex and impossible to write into a series of commands. That is why we consider the project as an experiment and the first step in the development of other methods of communicating with a computer. The tools being developed by Neuralink and Kernel allow us to expect that soon much more accurate and technologically advanced solutions will be available, which have a chance to change design methods.