Neural network model for recognition and classification of types of interactions in road traffic

The article presents neural network for recognition of driving strategies based on interactions between drivers in road traffic. It analyzes the architecture model implemented as a self-organizing map (SOM), consisting group networks radial basis functions (RBF). is training grounded biological foundations artificial networks, which set should consist exclusively input vectors; wherein algorithm adjusts itself network's weights to obtain consistent output vectors (i.e. make presenting sufficiently close result into same outputs). results using new generation developed by us, includes an adaptive learning reduce effect re-training (overfitting) and false recognition, well improve determination boundaries clusters. aim research outline structure that allows recognizing strategical characteristics can identify vehicles (their drivers) traffic behavioral patterns This paper considers characterize interaction dyads (drivers) moving show SOM RBF recognize classify types modelling analysis vehicle movement trajectories. Experimental demonstrate involving 400 iterations streaming data representing 500 possible simulated situations. novel drivers' behaviour classification five general classes: (1) competition strategy, (2) contest (3) evasion (4) compromise (5) active confrontation strategy. has demonstrated high rate concise clusterization similar strategies. key contribution this paper: it proposes Kohonen's Self-Organizing Map (SOM) detecting behaviours from - instead monitoring driver's specific activities.