Confined Space Mapping Module for In-Pipe Repair Robots

Default ARPA-E Project Image

Pittsburgh, Pennsylvania
Project Term:
11/04/2020 - 11/03/2023

Critical Need:

Legacy pipe malfunctions create operating risk and legal liability for utilities, negatively impact system owners’ financial performance, and are costly to gas consumers. Today, repairing legacy pipes involves excavating and replacing them, typically with high-density polyethylene pipe. Replacement costs range from $1-10 million per mile, depending on the pipe’s location (rural vs. urban), complexity of the excavation, and costs for restoring roads. Utilities also incur costs whenever gas service is disrupted for repairs. REPAIR seeks to eliminate the highest cost components, excavation and restoration, by rehabilitating pipes without their removal—in essence, by automatically constructing a new pipe within the old. To enable this effort, field technicians and engineers require 3-D maps of pipeline infrastructure, and REPAIR technologies also need 3D maps to record items such as anomalies, leaks, wall thickness, and results of coating deposition.

Project Innovation + Advantages:

Carnegie Mellon University (CMU) will develop a general-purpose mapping system that can integrate with virtually any mobile robot dedicated to pipe inspection and repairs. Confined spaces challenge map creation because they limit payload size. This not only affects the choice of sensor, but how its information is processed because of the space required to store edge computing. On top of that, confined spaces challenge the use of the sensors themselves; most sensors have a lower limit on sensing range, which is often violated in small spaces. Finally, pipe-confined spaces lack features, making mapping quite difficult. The team’s approach (1) fuses data from multiple sensory sources to accurately and reliably map the structure of a large-scale, distributed pipe structure while gathering information on its health, (2) creates a novel sensor payload specifically designed to operate in confined spaces to provide visual and geometric inspection, and (3) takes in and manipulates large-scale point cloud data in computationally efficient ways to improve visualization and the opportunity for modern techniques in AI to provide post-processing support.

Potential Impact:

REPAIR seeks to eliminate the highest pipe rehabilitation costs, excavation and restoration, by repairing pipes without their removal.


REPAIR projects should improve the sustainability of domestic natural gas distribution by economically rehabilitating legacy pipes.


REPAIR will produce 3D maps and data management/visualization tools that integrate geospatial data for leak testing, integrity/inspection data, coating deposition data, and locations of pipes and adjacent underground infrastructure.


REPAIR program innovations will accelerate legacy pipeline replacement while reducing cost to utilities and gas customers, potentially including the avoidance of backhoe strikes or other errors related to incorrect “one-call” markings.


ARPA-E Program Director:
Dr. Jack Lewnard
Project Contact:
Prof. Howie Choset
Press and General Inquiries Email:
Project Contact Email:

Related Projects

Release Date: