Spatial reasoning mechanism to enable automated adaptive trajectory planning in ground penetrating radar survey

Abstract

Lack of accurate and complete utility records is a root cause of utility strikes that occur on average once every minute in the United States, leading to hundreds of deaths and billions of dollars in financial losses. Ground-penetrating radar (GPR) has emerged as a promising tool to detect, locate, and measure underground pipes. Many algorithms have been devised to process GPR scans to determine pipe location and burying depth. Their accuracy depends on the relative angles between the GPR survey trajectory and the buried pipes. Perpendicular-to-pipe scanning yields the highest detectability, and along-pipe scanning yields the highest planimetric and depth accuracy. However, the challenge in practice is to maintain such ideal angles while not knowing the exact orientation of the pipes. This paper devises a novel spatial reasoning mechanism to enable the automation of GPR survey trajectory planning and adjustment based on ill-shaped and incomplete GPR signatures to achieve ideal angles in real time. The spatial reasoning mechanism provides trajectory adjustment suggestions based on the connection between the GPR signatures extracted from GPR scans and the relative angles between the GPR trajectory and underground pipes. The adjustment process continues until the GPR signatures from synthetic data under ideal angles and from field survey converge, and further adjustment no longer improves the results. Both indoor and field experiments have been conducted for validation. The results show that the newly developed method is capable of guiding the adjustment process in real time to achieve ideal angles and collect high-quality GPR data, leading to a more accurate estimation of pipe locations and burying depths. It has the great potential to support the operation of robotic unmanned ground vehicles (UGV) or unmanned aircraft system (UAS) to fully automate the GPR field survey.