Developing an unmanned aerial vehicle-based rapid mapping system for traffic accident investigation

Abstract

After a traffic accident, a diagram of the scene needs to be prepared as evidence in on-scene investigation. However, traditional methods to measure and map have their own defects. The aim of this study is the development of an unmanned aerial vehicle (UAV)-based mapping system to acquire scene diagrams rapidly. The camera calibration is explained in the study. This system provides approaches for lens distortion and perspective distortion elimination. A method for scene mapping from rectified images is also explained. The performance models for estimating the position error is developed and tested in different levels. The results show that the total root mean square error (RMSE) of the rectified image is 3.6 pixels, and the relative error of measurement in the rectified image can be controlled to –2.5%~2.5% when the UAV is at 11–112 m flight altitudes. This proposed method to draw scene diagrams is faster and easier than hand-drawing. It is indicated that the system provides a feasible method to obtain a scene diagram of a traffic accident.