UAV-Assisted Bridge Inspection: An End-to-End Operational Workflow Development and Testing
11th Annual COE Graduate Poster Presentation Competition
Student: Emmanuel A. Marfo (MS)
Advisor: Dr. Ali Karimoddini
- The US is home to over 615,000 bridges, with NC housing a little below 19,000 for which about 13% are rated as structurally deficient.
- Bridges, like any other structure, deteriorates with time and hence, requires frequent inspection and maintenance.
- E.g. The Mianus River bridge collapse on I-95 in 1983 - hanger assemblies failure.
- Conventional inspection practices, manned approach, have drawbacks like:
- Not safe (for both inspectors and nearby traffic).
- Not time efficient (FHWA approximates 1-3 days for an inspection).
- Not cost effective (FHWA quotes $2.7bn as bi-annual expenditure).
- Lane closure resulting to increased traffic.
Experimental results and recent literature confirms UAV’s tendency of:
- Ensuring safety on the sides of both the inspection team and the public.
- Reducing the total bridge inspection time.
- Reducing total expenditure per bridge inspection.
- Ensuring ease of access to hard-to-reach areas on the structures.
Therefore, definitive guidelines and resource tools for the deployment of UAVs in bridge inspections becomes essential in optimizing the entire process.
- To identify challenges associated with bridge inspections.
- To develop a guidance system that supports the identification of candidate bridges for UAV-assisted inspections.
- To develop an end-to-end operational workflow for UAV application in bridge inspection.
- To develop a crack detection system and resource tools for estimating and planning the inspection
- To develop a guidance system that supports the integration of data obtained by UAV into inspection reports.
This work is supported by the ACCESS Lab, NCAT and authorized by the North Carolina Department of Transportation under Master Agreement Contract: MA-2019-05
The proposed inspection procedure is divided into 3 sections sequentially as pre-inspection, inspection and post-inspection.
The level of abstraction in the workflow design takes into consideration technology transfer from research team to inspection teams.
Experimentally, 3 UAV platforms were selected based on experimental results, inspection requirements and objectives.
- In present UAV-assisted bridge inspection practices, the pilot and the inspector are different persons.
- With only the pilot having real-time view of the UAV feed, a room for deviation from the optimum can be said to have been created
- In effect, an inspector-in-the-loop concept is introduced. In this, the inspector has real-time view and access to the feed and hence can direct accurately the inspection and the flight in general.
- Wells, Jennifer, and Barritt Lovelace. Improving the quality of bridge inspections using unmanned aircraft systems (UAS). No. MN/RC 2018-26. 2018.
- Kim, B.; Cho, S. Automated Vision-Based Detection of Cracks on Concrete Surfaces Using a Deep Learning Technique. Sensors 2018, 18, 3452. https://doi.org/10.3390/s18103452
- 3D photorealistic models of the structure are created using the concept of Reality Modeling.
- The models can be used to measure deficiencies and can be annotated with inspection notes
- With this, great deal of time is saved since the inspector no longer need to create a photo log of the inspection
Vision-based algorithms can be taken advantage of during bridge inspections with UAVs. As such, real-time crack detection, crack width measurements and crack quantifications can be done easily thereby reducing the post inspection process requirements.
Applying the techniques of Augmented Reality to 3D reconstructed models of the inspected structure promises to ensure safety to the highest degree.
The inspection, crack detection and crack measurements can be conducted in office spaces.
This can greatly enhance the reporting and communication processes.
- It also creates a record of the bridge condition at a point in time so that as the bridge ages, deterioration rates can be observed.