An Algorithm Based on Random Forest for In-vehicle Navigation Pattern Recognition

In view of the problem of inaccurate navigation pattern recognition in complex driving environment, a navigation pattern recognition algorithm based on random forest is proposed. Firstly, the error source of navigation components is analyzed to determine the factors affecting the accuracy of different navigation patterns, as the characteristic vector of designing random forest models. Secondly, the data set is constructed by recording the data collected by the vehicle. The random forest model is trained with 70% random data in data set, and the recognition rate is verified with the remaining 30% data. The average recognition rate of this algorithm is 88%, which can meet the needs of the work.