University of British Columbia
Modernizing pavement condition assessment : a unified framework integrating light detection and ranging with computer vision to overcome index fragmentation
Abstract
dc:descriptionPavement condition assessment remains fragmented across transportation agencies, with inconsistent definitions, thresholds, and rating scales that limit comparability and decision consistency. Part of this inconsistency is also attributed to subjective assessment practices that rely on inspectors manually evaluating site conditions. To address these issues, this dissertation develops and validates a unified framework that combines lidar-based deformation measurement and vision-based crack measurement into a single standardized pavement condition index. The work progresses in four stages. First, a survey of 14 Canadian transportation agencies documents the current state of practice and confirms substantial index heterogeneity. Second, lidar methods are developed and validated for rutting, roughness (IRI), and pothole detection and quantification. These methods achieve practical agreement with references, including rut-depth deviation near 2.5 mm at operating thresholds, average IRI RMSE below 0.05 m/km, and pothole geometric error below 5.6%, while identifying practical minimum densities of approximately 230 ppsm for stable rutting/IRI and 205 ppsm for pothole reliability. Third, a top-down crack segmentation and metrology workflow is developed for near-nadir roadway imagery, achieving mean F1 = 0.696 and mean IoU = 0.692 at the operating threshold 𝛕* = 0.90 across a bimodal per-image performance distribution, and producing physically interpretable crack metrics (density, width, and coverage). Fourth, these deformation and cracking outputs are integrated through severity-transfer functions, calibrated weighting, and section-level fusion to produce a Unified Pavement Condition Index. The full implementation is demonstrated on a 1.5 km segment of Highway 97 through downtown Kelowna (15 sections). The computed corridor condition is PCI(100) = 83.4 (PCI⁽⁵⁾ = 1.66). Validation against independent windshield ratings shows meaningful agreement with expert judgment (Spearman 𝛒 = 0.81, MAE = 0.30 on the 1-5 scale, ICC(2,1) = 0.85), with all sections within ± 1 condition class. The resulting framework provides a repeatable, implementable basis for standardized pavement condition assessment using complementary sensing technologies.
Degree
thesis:*- Name thesis:degree_name
- Doctor of Philosophy - PhD
- Level thesis:degree_level
- doctoral
- Discipline thesis:degree_discipline
- Civil Engineering
- Grantor dc:publisher
- University of British Columbia
- Year dc:date
- 2026
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Faisal, Ali J.
- Contributors dc:contributor
-
- Gargoum, Suliman
Rights
dc:rights- Statement dc:rights
-
- Attribution-NonCommercial-NoDerivatives 4.0 International
- Language dc:language
- eng
Identifiers
dc:identifier.*- Handle dc:identifier
- http://hdl.handle.net/2429/95075