KAMAZ implements the module of voice actuation from CST

"KAMAZ" implements the module of voice actuation from Group of CST in an onboard information system. Voice actuation in the truck in Russian is implemented for the first time. A project objective — not only to increase comfort level for drivers who will be able not to be distracted by clicking of buttons during driving, but also to increase traffic safety.

The first stage of the project is handed successfully over by the fall of 2016, is going to release cars with such module in 2017.

Features

Using this system the truck driver will be able "without hands" to use a multimedia and navigation complex. Voice actuation is activated by the button on a wheel. It is not necessary to remember any commands – management is implemented on the basis of speech recognition. With its help the driver can:

• Manage navigation
• Manage multimedia (to switch songs on a disk, folders, to change radio)
• Manage the telephone directory
• Make primary diagnostics of the car (for example, to learn accumulator charge level), etc.

The module from CST is not limited in subjects of requests in any way – developers of an onboard information system can build in new subjects there. Thanks to the built-in system of voice interaction the module will be able to communicate by the synthesized voice with the driver: specify wishes, confirm commands and other.

Technical features

Basis of the module of voice actuation from CST – sensing technologies of a spontaneous (unprepared) speech and its synthesis. Recognition is responsible for understanding of the driver, and synthesis for communication with it. Interaction is performed on the basis of elements of artificial intelligence: if the request is formulated in a form, unclear for a system, she will be able to specify it, having offered options.

At the beginning of the project developers were faced by an ambitious task: create the tool which will be able to work in any place where there can be "KamAZ" – and it is any point of Russia where can not be the Internet and cellular communication. Therefore a system is implemented so that recognition and speech synthesis were performed without connection to worldwide network.

One more call standing before R&D-командой CST consisted in that a system could understand the speech during the movement of the truck in the conditions of noise (for example, at open windows in a cabin). For quality improvement of recognition in the movement the optimal arrangement of the microphone was selected from a cabin, different types of noise were studied and algorithms of recognition and a shumoochistka were configured. That it is the is most complete to reproduce an acoustic situation, experiments with record of voice commands different microphones in different models of the KamAZ cars were made and under different traffic conditions. As a result very high rates of accuracy of recognition were reached.

Results, perspectives

The main objectives of the project are carried out: the module will recognize the driver's voice during the movement without connection to network and answers it. On the next stage of works the Group of CST and "KamAZ" are going to finish the device during trial operation and in 2017 to release the first "hearing" and "speaking" trucks.