Abstract:The forward collision warning system (FCWS) has often been facing such practical problems as false warning or warning omission and poor acceptance by drivers,so an adaptive safety assistance algorithm for longitudinal collision avoidance is presented and a driver following habit characteristic model based on back propagation(BP) neural network is established in this paper.The inputs of this closed-loop model include the velocity and acceleration of preceding car,the velocity of following car,the environmental luminance,the tire-road friction coefficient,and the emergency braking times,and using adaptive learning rate method based on momentum gradient descent to train this network model,the desired acceleration of following car can be predicted.The predictive performance of this BP network model is improved by introducing the idea of feature clustering algorithm.Meanwhile,three typical driving groups,mainly composed of radical,cautious,and novice groups,are designed,whose behavior is characterized by braking depth,desired reciprocal of TTC and emergency response time.By means of learning the driving behavior of different groups in the process of steady car-following,a nonlinear mapping relationship of knowledge base is established,which makes it possible to predict the corresponding braking behaviors,and accordingly the differential pre-warning can be realized,which will effectively reduce the probability of false or not timely warning.Detailed simulations show that the improved BP network can reduce the possibility of local optimum,and it also improves the network's convergence rate and the accuracy of prediction.Further,the downward trend of braking curve of radical and novice groups is steeper than that of cautious groups,and in addition,the car-following distance of radical groups is relatively shorter,while for the novice groups it is relatively longer to compensate for the longer response distance.From the simulation results,it shows that for different driving groups,the proposed model shows good adaptability.

Car Following Model Under Emergency Evacuation Situation Based on the BP Artificial Neural Network

Exploration of Car-Following Model in Emergency Evacuation Situations

Differences in Driving Characteristics Between Normal and Emergency Situations and Model of Car-Following Behavior

Study on the Prediction Meth of Automobile Braking Distance Based on Artificial Neural Network

Car-Following Modeling Incorporating Driving Memory Based on Autoencoder and Long Short-Term Memory Neural Networks

Car-following model based on artificial potential field

Driving Behaviour Style Study with a Hybrid Deep Learning Framework Based on GPS Data

Driver Car-Following Behavior Modeling Using Neural Network Based on Real Traffic Experimental Data

Research on Short-Term Driver Following Habits Based on GA-BP Neural Network

Network based on real traffic experimental data

Application of Artificial Neural Network and Particle Swarm Optimization in Car-Following Model

A Car Driver Assistance Model Based on Neural Network and Hidden Markov Chain

Safety Assistance Algorithm for Longitudinal Collision Avoidance Based on BP Neural Network

A Study on Prediction of Vehicle Critical Follow Distance Based on Driver's Behavior by Using BP Neural Network

ADAS Simulation Result Dataset Processing Based on Improved BP Neural Network

Comparison among Driving State Prediction Models for Car-Following Condition Based on EEG and Driving Features.

Research on Passenger Evacuation Behavior in Civil Aircraft Demonstration Experiments Based on Neural Networks and Modeling

Vehicle Driving Risk Prediction Model by Reverse Artificial Intelligence Neural Network

Composite Driver Car-Following Model Based on Neural Network Approach

A High-Precision Car-Following Model with Automatic Parameter Optimization and Cross-Dataset Adaptability

Safety Evaluation of Traffic Accident Scene Based on Artificial Neural Network