Research on Intelligent Vehicle Regulation Algorithm Based on Driving Intent Recognition

In this paper, a multimodal fusion driver intention prediction system is proposed based on the driver lane change intention reasoning framework. The system utilizes the improved Attention-BiLSTM neural network model for intention prediction by analyzing the driver's behavior, traffic environment and vehicle physical state. By preprocessing and analyzing the data from the Brain4cars database, the driver's head pose and driving scene features are extracted, thus providing rich input information for the model. The experimental results show that the model proposed in this paper achieves more than 92% accuracy in recognizing the driver's intention to change lanes, which is significantly better than the traditional support vector machine and LSTM models. This research provides the theoretical foundation and technical support for the realization of intelligent driver assistance systems, and has important application prospects.