Vehicle Acceleration Prediction using Specific Road Curvature Points

Abstract

In the work vehicle acceleration prediction issue is discussed. Three types of parameters are used for prediction system input: CAN-bus parameters - speed and curvature, derived speed parameters and newly offered specific curve point parameters, denoting changes in a curve. The real road data was used for predictions. Road curvature segments were divided into single and S-type curves. Acceleration was predicted using artificial neural networks and look-up table. The look-up table method showed the best results with newly offered specific curve parameters.