Mapping the Underworld of Buried Utilities - A Hybrid Sensing Approach

Purdue University, West Lafayette IN

Investigators

Abstract

This grant will pioneer a novel hybrid sensing approach to accurately map underground utilities using both sensor and non-sensor data. For the massive underground utilities, knowing their locations, dimensions, spatial configurations, and materials is critical to effectively maintaining, restoring, and upgrading them to provide essential services to the society: water, electricity, gas, and the Internet. Lack of accurate utility location information is the main reason for over six million incidents of excavation damage to utilities every year, causing injuries, fatalities, and property damage, on the order of billions of dollars. Current sensor-based utility mapping methods are less effective in the utility-congested urban environment where many utility lines occupy a tight space, causing sensor signal interferences and consequent erroneous utility information. The anticipated approach will create a method to supplement textual data from utility regulations to digital sensor data, enabling automated and more accurate mapping of underground utilities. With accurate utility information, 22% of the excavation-related incidents could be avoided. Therefore, results from this research will benefit society through cost savings, reduced hazards to citizens, consistency in services supplied by critical underground lifelines, and improved sustainability of urban communities. Research outcomes will be integrated into engineering curriculum, K-12 education, and industry workshops to promote subsurface utility engineering as a specialization area. The objective of this research is to create a hybrid, multi-sensing method to automate the process of interpreting and fusing heterogeneous utility data to generate accurate utility maps. This novel approach is expected to shift the current paradigm in mapping and labeling underground utilities in utility-congested urban areas by devising methods that exploit utility specifications and plans as a "virtual sensor." This approach will provide contextual information for underground utilities. This information will be integrated with information from the processing of ground penetrating radar(GPR) scans. The research team will create algorithms to automatically detect multiple and transformed signatures from underground utilities created with the GPR scans, estimate utility locations, dimensions, and spatial configurations, and recognize pipe materials. Natural language processing algorithms will be devised to mine the textual data in utility specifications and plans. These algorithms will enable extraction of spatial rules constraining utilities and their surroundings, which will be fed into a spatial reasoning algorithm. This use of utility specifications and plans for this purpose is revolutionary. A novel framework for utility data/information fusing will be created to address spatial inference and data correspondence computation problems in utility mapping applications that are as of yet untapped. The credibility of the system and methods will be validated at each research phase through experiments.