The new model developed by the ISSAI scientists can determine the location of miners underground based on Wi-Fi signals. Researchers from the Institute of Smart Technologies and Artificial Intelligence believe that their new development will help miners feel safer underground.
While working on the project, the ISSAI team visited the Donskoy mine in Khromtau and the Aktobe Ferroalloy Plant. Today, the project team is collecting the necessary data. Once all the data are collected, the team will start building the machine learning algorithm.
“The model will accumulate data by receiving them in the form of available Wi-Fi spots signals, as well as in the form of IMU (Inertial Measurement Unit) sensors. The most common examples of IMU sensors are built-in pedometers in smartphones. These pedometers have three sensors that can determine the direction of the acceleration of a person and the acceleration of the Earth, and using these parameters, IMU sensors determine the location of a person and magnetic field of the Earth. For example, if Wi-Fi networks are available in certain places, we can determine the location of a person, but if this person goes off the Wi-Fi grid, we still can locate him, using IMU sensors to determine the direction and distance,” said ISSAI data analyst, Mukhamet Nurpeissov.
To develop the model, ISSAI researchers will train deep neural networks using the Nvidia DGX-2 supercomputer. The project team will train the model to locate a person based on incoming signals received from both Wi-Fi and IMU sensors. The developers are confident that in the future, if necessary, the model can be adapted and used to locate people in any public place.