Justification of a set of informative features for recognition and characterization of current situations when driving a vehicle. 2019-01-0100
The quality of recognition of the road situation in which the driver is, depends on the quality of the features used by the classification system. The main characteristic of the feature is its informativeness. Moreover, informativeness is a relative value, which is expressed in different information values of a separate attribute for the classification of various road situations. Also, informative character may depend on the type of deciding rules in the classification procedure.
Modern vehicles are equipped with powerful diagnostic systems, control of current traffic parameters and input signals. Therefore, at the disposal of the systems of pattern recognition of train situations there is a set of features that can count several hundred members. Using the entire set of attributes may not be appropriate for the following reasons. Each sign needs a separate information channel for its control, which complicates the interface part of the system being developed and leads to its growth. The second factor is the use of computing resources onboard computers, which at the moment are at least powerful enough, but it is desirable to have the maximum speed of the system on the vehicle, which is impossible when processing and analyzing such a wide range of data.
Thus, the task of evaluating and choosing informative features when recognizing road situations is relevant and will allow to improve the efficiency of on-board system of image recognition system.
Citation: Gritsuk, I. and Horobchenko, O., "Justification of a set of informative features for recognition and characterization of current situations when driving a vehicle.," SAE Technical Paper 2019-01-0100, 2019. Download Citation
Igor V. Gritsuk, Oleksandr Horobchenko
Kherson State Maritime Academy, State University of Infrastructure and T
SAE Connected and Automated Vehicle Conference Israel