Denso company, specializing in the production of automotive components, and the company Toshiba, which occupies a leading position in the microelectronics market, signed an agreement on joint development of the technology of artificial intelligence, called Deep Neural Network-Intellectual Property (DNN-IP). It will be used in pattern recognition systems, the two companies independently developed and designed for automotive driver assistance systems and autonomous driving.
The forecast is that the deep neural networks implemented in the DNN-IP and simulate the human brain will be able to recognize patterns no worse than it does men. Confident pattern recognition, allowing to assess the traffic situation with signs, markings and obstacles of all kinds, is a key element of autonomous driving systems.
In contrast to the samples of recognition on the basis of comparison, in which all objects must be described and entered into the system’s memory in advance, detection using DNN-IP enables a computer to learn and to extract the characteristics of the objects necessary for the recognition, by yourself. This provides a higher recognition accuracy and extended range of recognizable objects.
Source: of Denso