Intelligent Vehicle Maneuver Detection Based on Synthetic Gyroscope Data Calculated from GPS

Accurate detection and classification of driving maneuvers play a crucial role in understanding driver behavior, enhancing road safety, and preventing accidents. This study proposes an algorithm for maneuver detection based on GPS data. The method involves preprocessing raw GPS data, transforming coordinates into a local reference frame, and estimating gyroscope data using Krylov-Euler angles. Maneuvers are identified by detecting significant changes in angular velocity, and their boundaries are determined dynamically. A template-based approach is used to classify maneuvers into left turns, right turns, and U-turns, with templates constructed based on the total angle change during the maneuver. To evaluate the proposed method, we randomly selected 10 drivers who completed 54 trips, totaling 11 hours of driving, from the DriverSVT Dataset. The dataset includes trips recorded in different geographical settings. It has segments from urban areas, expressways, rural roads, and residential areas such as backyards. This diversity allows the algorithm to be tested in a variety of real-world driving conditions. The proposed method provides a robust way to analyze driving behavior. Experimental results demonstrate the effectiveness of the approach in accurately segmenting trips and classifying maneuvers.