A Spherical Pipeline Inspection Gauge (PIG) for In-Line Navigation & Defect Detection

On June 5, 2024, Calgary’s Bearspaw South Feeder pipe, the largest feeder main in Calgary’s network, suffered a catastrophic burst. [1] Within a month, the city had reduced water use to 425 million litres, a 29% reduction. [2] On December 30, 2025, the same feeder main broke again, forcing city-wide water restrictions and emergency repairs. [3]

These pipelines are essential to life, delivering clean drinking water, supporting public health, enabling emergency services, and sustaining economic activity. [4][5]

Deteriorating pipeline infrastructure has become a prevalent problem in the 21st century, with multiple municipalities facing constant repairs, flooding, and service outages. [6] Nearly 20% of water pipeline infrastructure in North America is beyond its expected life, increasing the risk of failure. [7] Current inspection methods can be expensive, difficult to deploy, or impractical, leaving cities reacting to failures instead of preventing them. [8]

This project proposes a smart Pipeline Inspection Gauge (PIG) to combat the issues posed by current technologies. It demonstrates the possibility of being deployed into 8’’+ PVC, cast iron, asbestos-cement, PCCP,  etc., municipal water pipes.

This project follows three objectives:

Navigation & Practicality: Construct an easy-to-deploy system that could move in a controlled, repeatable manner through straight sections, 45° bends, 90° bends, and T-junctions.

Accuracy: Obtain a mean depth error of less than 2 cm in controlled tests and generate inspection output clear enough to reveal crack-like surface defects

Cost Effectiveness: develop a prototype that would remain under C$500, and use common components on a small scale.

Mechanical Design

A comprehensive literature review, alongside conversations with industry experts, was conducted to analyze state-of-the-art PIG designs. Initial ideas included a biologically inspired “worm” robot and a helical screw-like rotating robot. However, these designs introduced limitations such as mechanical complexity and slower movement. To address these shortcomings, a compact spherical chassis was selected for the final design.

Robot Architecture: The compact spherical chassis of the robot improves its insertion and retrieval capabilities while reducing snag points in bends, joints or transitions attributed to its ability to rotate freely.

Drivetrain, Traction, and Torque: The drivetrain uses an internal hamster-ball-style drive mechanism. Two powered wheels, pressing against the inside of the spherical shell, rotate and create friction that causes the outer shell to roll through the pipe. Passive stabilization wheels support the internal frame. To improve traction, the wheels were redesigned with a rounded ridge profile and a PLA hub + TPU tread. A 50:1 gear ratio was chosen to prioritize torque and reliable startup/traversal over raw speed.

Sensing Mechanism

After evaluating LiDAR and sonar, the sensing system moved from emission-based sensing to a computer-vision approach. To estimate depth in 3D space, a stereo vision system was utilized, by positioning two cameras (like human eyes) to determine depth anomalies such as cracks:

(a) Stereo Image Capture: Two synchronized lenses capture the same pipe from slightly different viewpoints.

(b) Left and Right View Separation: The side-by-side frame is split into left and right images for stereo comparison.

(c) Depth Map Generation: The horizontal shift between views is converted into a depth map.=

(d) Image Alignment After Calibration: Calibration aligns matching features onto the same horizontal rows.

In addition, an RGB camera captures images of the pipe wall. This combines human-reviewable imagery with 3D reconstruction for identifying cracks and visible surface defects.

Testbed & Protocol

Before testing began, a realistic testbed was developed, along with a controlled protocol to ensure the results were accurate. Tests a-c were tested through 5 trials, with all runs having the same initial configuration and battery level (approx full). Each run captured various sensor outputs (IMU, motors, stereo camera), alongside empirical results such as traversal times and completion rates. 9-inch 3D-printed pipes–including straight, 45°, 90°, and T-junctions–were selected to mimic the complex and unpredictable geometry of municipal piping systems. Defects were stimulated by designing cracks in various diameters, latitudinally and longitudinally across each unique pipe section. Tests were conducted in both dry and semi-wet conditions (using a garden hose for water flow).

a. Mobility & Stability

To achieve the robots' first objective–the ability to effectively navigate through complicated pipe geometries–mobility was rigorously tested. Trials showed that performance had a general linear speed decrease as pipe geometries became more complex, due to the friction coefficient dropping. The robot completed 5/5 straight trials and 5/5 trials through 45° bends, with average speeds of 0.075 m/s and 0.055 m/s, respectively. Performance began to decrease as the robot completed 90° bends and T-Junctions, with 4/5 at 0.038 m/s, and 3/5 trials at 0.025 m/s, respectively. As the spherical chassis risked the chance of drifting out of place, an IMU based PID control system was implemented. Proportional heading control was incrementally tweaked using multiple gain values. Results show that lower gain corrected disturbances slowly, while higher gains cause large oscillations. The final gain value–a balance in between–remained the best compromise as it reduced heading error swiftly, without causing unstable motion.

b. Stereo Output

To achieve the robot’s second objective–to accurately detect cracks in a pipe's inner lining–stereo output trials were conducted through a variety of unique tests. After each video frame was converted into depth maps and point cloud reconstructions, it was found that the system produced a usable reconstruction in 89% of trials, exhibiting a mean depth error of 0.63cm and a best-case error of 0.28cm on cracks 2mm in width and above. Point-cloud completeness had reached a maximum accuracy of 83% in a working distance range of 15-40 cm.

c. Practicality/Cost/Runtime/Pipe-fit

To achieve the robot's third objective and further investigate the first, cost calculations were conducted, alongside theoretical estimates to determine practicality, runtime, and pipe-fit. The final prototype cost is approximately C$362, significantly surpassing the initial goal of being just under C$500. Data storage is programmed to be held locally, and when retrieved, it can be processed. Only one operator would need to handle this robot, which requires 3-5 minutes of setup time and an estimated continuous in-line runtime of 45-55 minutes. As the spherical shell’s diameter is 6.25 inches, pipe-fit analysis calculations show that the robot is best suited for 8-12 inch mains(could be higher if the robot's size is increased).

Conclusion

While this research remains preliminary, this robot has demonstrated the proof of concept for a low-cost, compact, first-pass inspection robot for 8-12 inch municipal water distribution pipes. While state-of-the-art counterparts often require expensive equipment and tethered deployment, this robot serves as a practical alternative.

Key findings

The unique combination of vision in a ball-like design provides a strong balance between mobility, simplicity, and inspection output.

Because of its design, the spherical shell makes it so it essentially removes any room for error from snag points during traversal. With the addition of IMU based PID stabilization, junctions can be inspected with ease due to active stabilization

The stereo vision systems add extra value in that not only does it provide a comprehensive depth map for users to analyze, but also high FPS RGB footage to follow that for highly accurate review–cracks can now be inspected with both visual and geometric information

Limitations

The prototype has not yet been tested under municipal pipe pressure conditions

Battery life limits the inspection duration

The current robot is suited for 8-12 inch pipes, with miniaturization needed for smaller pipes

Further Use Cases

The robot's use case doesn’t need to be just pipe inspection. Domains like search and rescue, hazardous-environment monitoring, and confined-space infrastructure inspection can easily be added to the functionalities of this system. Its robustness allows it to traverse multiple environments at ease.

Planned Future Work

Future work for the robot includes running it through controlled pressure-rated pipe validation tests. An improved test bed would use a hydrostatic loop to control water pressure at 50, 75, and 100 PSI, evaluating the robot's waterproofing, changes to shell formation, traction to pipe, and sensor performance. Internally, the following refinements are proposed:

Attach a machined O-ring for durable waterproofing

Use a transparent Hard-Coated Polycarbonate shell to increase durability and scratch resistance

Utilize sensor fusion techniques, such as combining acoustic & electromagnetic techniques alongside what was developed for a highly accurate defect detection rate.

This project would not have been possible without the support of many individuals. First, I would like to thank my school science fair coordinator, Mr. Bobby Lahoda, for supporting the submission of my project to this prestigious event. I would also like to acknowledge Luke Ryan for his continuous guidance in the design of my project and for providing feedback throughout the iterative process. Finally, I would like to thank my parents for supporting me throughout this journey and always believing in my goals.

References

The City of Calgary. (n.d.). Background: 2024 Bearspaw South Feeder Main repairs. https://www.calgary.ca/emergencies/feeder-main-repair/faq.html

City of Calgary Newsroom. (2024, July 1). Update July 1: Critical water main break affecting city-wide water usage. https://newsroom.calgary.ca/update-july-1-critical-water-main-break-affecting-city-wide-water-usage/

The City of Calgary. (n.d.). Repair updates and images. https://www.calgary.ca/emergencies/feeder-main-repair/water-main-break-updates.html

Health Canada. (2025, December 5). Drinking water and health: Overview. https://www.canada.ca/en/health-canada/services/environment/drinking-water.html

American Water Works Association. (2012). Buried no longer: Confronting America’s water infrastructure challenge. https://ipp-inc.com/wp-content/uploads/2020/01/American-Water-Works.pdf

Barfuss, S. L. (2023). Water main break rates in the USA and Canada: A comprehensive study. Utah State University. https://doi.org/10.26077/6jhf-md25

Landers, J. (2024, April 22). Breaking water mains present US, Canada with $452B problem. American Society of Civil Engineers. https://www.asce.org/publications-and-news/civil-engineering-source/civil-engineering-magazine/article/2024/04/breaking-water-mains-present-us-canada-with-%24452b-problem

Liu, Z., & Kleiner, Y. (2013). State of the art review of inspection technologies for condition assessment of water pipes. Measurement, 46(1), 1–15. https://doi.org/10.1016/j.measurement.2012.05.032

Bradski, G., & Kaehler, A. (2008). Learning OpenCV: Computer vision with the OpenCV library. O'Reilly Media.

Chung, J., & Kurup, A. (2021). A review on various types of in-pipe inspection robot. International Journal of Mechanical Sciences. https://doi.org/10.1016/j.ijmecsci.2021.106783

The City of Calgary. (2026, January 5). Repair progress — Jan. 5, 2026 [Photograph]. In Repair updates and images.

Digi-Key Electronics. (n.d.). TB6612FNG motor driver IC [Product page]. Retrieved March 4, 2026, from https://www.digikey.com/en/products/detail/toshiba-semiconductor-and-storage/TB6612FNG-C-8-EL/1027355

Folkman, S. (2018, March). Water main break rates in the USA and Canada: A comprehensive study. Utah State University, Buried Structures Laboratory. https://digitalcommons.usu.edu/cgi/viewcontent.cgi?article=1109&context=mae_facpub

Global News. (2024, June 6). A view of the aftermath of a water main break in northwest Calgary's Montgomery neighbourhood [Photograph]. In P. Heidenreich & P. Tran, Calgary activates Municipal Emergency Plan after water main break spurs boil water advisory. Global News.

Hartley, R., & Zisserman, A. (2004). Multiple view geometry in computer vision (2nd ed.). Cambridge University Press.

Hirschmüller, H. (2008). Stereo processing by semiglobal matching and mutual information. IEEE Transactions on Pattern Analysis and Machine Intelligence, 30(2), 328–341. https://doi.org/10.1109/TPAMI.2007.1166

Kakogawa, A., & Ma, S. (2018). Design of a multilink-articulated wheeled pipeline inspection robot using only passive elastic joints. Advanced Robotics, 32(1), 37–50. https://doi.org/10.1080/01691864.2017.1393927

Krause, D. (2025, December 31). Workers watch as a backhoe begins unearthing the newest Bearspaw South feeder main break in northwest Calgary [Photograph]. LiveWire Calgary.

Kwon, Y.-S., & Yi, B.-J. (2012). Design and motion planning of a two-module collaborative indoor pipeline inspection robot. IEEE Transactions on Robotics, 28(3), 681–696. https://doi.org/10.1109/TRO.2012.2183050

Liu, Z., et al. (2022). A review: Technological trends and development direction of pipeline robot systems. Journal of Intelligent & Robotic Systems, 105, 11. https://doi.org/10.1007/s10846-022-01669-2

National Research Council. (2006). Drinking water distribution systems: Assessing and reducing risks. The National Academies Press. https://doi.org/10.17226/11728

OpenCV. (n.d.). Camera calibration and 3D reconstruction (calib3d module documentation). Retrieved March 4, 2026, from https://docs.opencv.org/4.x/d9/d0c/group__calib3d.html

OpenCV. (n.d.). Depth map from stereo images (Python tutorial). Retrieved March 4, 2026, from https://docs.opencv.org/4.x/dd/d53/tutorial_py_depthmap.html

Pololu Corporation. (n.d.). Micro metal gearmotors [Product category]. Retrieved March 4, 2026, from https://www.pololu.com/category/60/micro-metal-gearmotors

Roh, S., & Choi, H. R. (2005). Differential-drive in-pipe robot for moving inside urban gas pipelines. IEEE Transactions on Robotics, 21(1), 1–17. https://doi.org/10.1109/TRO.2004.838000

Scharstein, D., & Szeliski, R. (2002). A taxonomy and evaluation of dense two-frame stereo correspondence algorithms. International Journal of Computer Vision, 47(1–3), 7–42. https://doi.org/10.1023/A:1014573219977

Toshiba Electronic Devices & Storage Corporation. (n.d.). TB6612FNG dual DC motor driver IC [Datasheet]. Retrieved March 4, 2026, from https://toshiba.semicon-storage.com/

U.S. Environmental Protection Agency. (n.d.). Drinking water distribution systems. Retrieved March 4, 2026, from https://www.epa.gov/dwreginfo

Xylem. (n.d.). SmartBall® in-line leak detection [Product page]. Retrieved March 4, 2026, from https://www.xylem.com/en-us/brand/pure-technologies/Products-and-Services/Leak-Detection/SmartBall/

Zhang, Z. (2000). A flexible new technique for camera calibration. IEEE Transactions on Pattern Analysis and Machine Intelligence, 22(11), 1330–1334. https://doi.org/10.1109/34.888718