Development of an autonomous robotic system for terrain mapping

Abstract

This paper describes the development of an autonomous mapping system (AMS) for basic terrain mapping and quantifies it's performance relative to conventional survey instruments. The AMS platform is an off-the-shelf robotic chassis equipped with wheel encoders, IMU and a survey-grade GNSS RTK receiver. The AMS is autonomously controlled by the Robot Operating System (ROS) using open-source and purpose-written software. The AMS has been used to produce a terrain map of a test site, which has been compared to terrain maps created by conventional survey methods. The results show that the AMS can autonomously map the test site in 13 minutes with an accuracy of 12-14 mm RMSE. The AMS has better mapping accuracy than manual continuous methods using a Total Station (33-51 mm) or GNSS (47-75 mm), and is comparable to UAV photogrammetry (14 mm) and Total Station scanning (15-16 mm).