Convolutional neural networks of active railway safety system with braking dynamics prediction
this study is dedicated to solve a collision prevention task by changing speed of the train by Convolutional Neural Network (CNN) application. System proposed in this paper is a part of the big research aimed at system, for the safe transportation process development. Different objects, such as car or a human, are important parameters in road circumstances analysing, collision possibility calculating and decision making for the accident prevention. System structure for collision possibility detection and minimization by braking dynamics prediction and CNN algorithm with training for object recognition tasks is proposed in this paper. The experimental results validate the effectiveness of proposed algorithm. Experiments show that CNN correctly defines all the objects after training is done.