High accuracy of the driver behavior prediction model is beneficial to driver assistant system and fully autonomous cars. My research proposes a lane changing prediction model based on the combined method of Supporting Vector Machine (SVM) and Artificial Neural Network (ANN) at highway lane drops.
The vehicle trajectory data are from Next Generation Simulation (NGSIM) data set on U.S. Highway 101, Interstate 80 and Tsinghua University. Different classifiers are adopted and compared to predict the feasibility and suitability to change lane under certain environmental conditions.
The environment data under consideration include speed difference, vehicle gap, and the positions. The best performance is the proposed combined model with improved accuracy, demonstrating the effectiveness of the proposed method and superior performance compared to other methods.