Systems based on artificial intelligence and neural networks can be adapted to perform a variety of tasks from complex matematicheskih computing to the recognition of objects and even video editing. Recently A group of Russian scientists developed a special device that will allow blind people to re-Orient themselves in space. The device is called, SOME — a system of multichannel information processing.
Responsible for the development of the company “Business Bureau” and, according to the newspaper “Izvestia”, is a gadget to analyze the surrounding space with the aid of the received video and audio signals. The device consists of two parts: the first is all the necessary sensors to record information like sensors, microphones and cameras, and the second is actually the platform itself is based on AI.
The principle of operation is as follows: having received the necessary environmental data, the system transfers them to the cloud, where the neural network analyzes the information and then the result is sent back to the user, where it is converted to voice or tactile message, or on the basis of constructed virtual image (albeit in a rather primitive form). The last option available with the ocular implant.
“System SOME will work with modern bionic implants — this is a feature of the device. I really hope that all conceived will be realized when the product will go into the series.” — said Denis Kuleshov, one of the leaders of the project.
Today, the system SOMSI is a digital analogue of the retina, but the processing of pictures in all details is not enough power and technology. To recognize major schematic outlines of the objects in the gadget is quite capable. Have SOME and another function: the definition of the distance to objects by means of ultrasound. The user will need to hold down just one button and the network will tell him where and what the subject is. At the moment is the improvement of device performance and market entry is planned in the second half of 2019. As reported by the engineer of the project V. Cheprakov,
“Our neural network is a very complex structure, which requires long and careful training. To recognize one thing, you need about 10,000 pictures in different angles. We train your network. While she recognizes the basic things, such as person, car or phone, but still does not distinguish between different persons or vehicles.”
In plans of developers is also included to teach the neural network to recognize faces of particular people and to transfer part of the program in the device itself, so it could work in the absence of the Internet.
This and other news you can discuss in our chat in Telegram.