The invention relates to the technical field of
vehicle driving safety, in particular to an ice and
snow covered pavement vehicle lane changing risk prediction method based on driving intention recognition, which comprises the following steps: collecting ice and
snow covered pavement multi-source heterogeneous driving data to recognize a lane changing intention; in the execution stage, based on real-time
kinematics and surrounding interaction information, a correction collision
time index of an ice and
snow adhesion coefficient correction factor is calculated and introduced, and a distance deviation index of an actual parking
sight distance is combined; the two indexes are mapped into
risk exposure and severity quantized values, and a comprehensive index is obtained through
fault tree analysis and divided into high, medium and low risk levels. According to the method, intention stage multi-dimensional features serve as input, graded indexes serve as labels, a
dynamic prediction model is trained and constructed through a
gradient boosting decision tree, early graded prediction of the ice and snow pavement lane changing risk is achieved, and
risk quantification accuracy and scene adaptability are improved.