Can Drones Survey Tucson Pavement Cracks and Rank Repairs?

Can a drone survey Tucson pavement cracks as well as a walking crew?

Tucson's paved city road network is far too large for a visual survey to keep pace with the cracks. According to the Arizona Daily Star, the city maintains 1,816 miles of paved roads, and according to KVOA the city has told the public that 70% of roadways are in poor or failed condition. A walking survey of that much pavement has to be sampled or stretched across seasons, and that is how a repair list ends up a season behind the streets it is meant to rank.

The real question is not whether a drone can see a crack. It is whether the images give your team a repeatable condition score that can be ranked. Tucson already rates its streets on an Overall Condition Index from 0 for failed to 100 for excellent, based on a visual survey, according to the Arizona Daily Star reporting. A drone survey can feed that same index, which keeps your repairs ranking comparable with earlier years and shows whether a crack is growing or just visible.

How does a drone pavement survey work on city streets?

A drone flies a set pattern over the road and captures overlapping photographs. Photogrammetry stitches those images into a surface model, and GPS tags tie each defect to a location, so a crew can find the same crack on a later pass. Case studies comparing drone and manual surveys report reduced field time and increased repeatability, and the GPS tie is what makes the repeat measurement useful.

The strongest state-level evidence comes from a Massachusetts UAS study. Its Volume 2 Task B report judged photogrammetry the most promising crack-detection approach paired with drones, and a U-Net deep learning model performed well on a small training set. The study used airport runway images, not city streets, so read it as method evidence rather than a Tucson result. Its best balance of efficiency and accuracy came from flying about 50 feet above the surface.

Measurement quality depends on flight height, lighting, image overlap and the training data behind the crack detector. Ask any provider whether the method was checked against AASHTO R 85, which covers quantifying cracks from automated images, and AASHTO R 86, which covers collecting pavement surface images. A provider who cannot answer that is describing a demo, not a survey method.

How accurate is a drone crack survey compared with a person?

Accuracy depends on the crack type, and no single number covers every distress. An Indonesian study in the journal Transportation Engineering, published in 2023, found drone distress measurements matched manual field measurements at 75.72% accuracy for edge cracking, the lowest category, while joint reflective cracking scored highest. The same study's drone-derived pavement condition index correlated with the manual index at r squared of 0.86, which is encouraging for ranking but still leaves room for error on individual streets.

Manual surveys are not a clean baseline. An Orange County Transportation Authority prequalification study, presented as a 2015 conference paper, found the pavement condition index standard deviation was about 2.8 at PCI 90 and about 18.1 at PCI 35. Only 3 of 21 inspectors met criteria in the 2013 round. Two inspectors can rate the same street very differently, and the spread is widest on the worst streets, which is where your list most needs to be right.

Because neither method is a perfect truth, the practical test is agreement on your own streets. Pick a sample across crack types and street classes, score it by hand with your most experienced inspectors, then compare the drone output. Expect the weakest agreement on edge cracking, the category where the Indonesian study scored lowest.

What does Tucson's pavement repair season look like?

Tucson's calendar sets the rhythm. According to an archived National Weather Service page, the monsoon season runs June 15 through September 30, a fixed window the NWS adopted in 2008. A spring survey lands in the gap before the summer storms, which gives you a baseline, and a second pass after September 30 shows what changed through the season.

No citable source ties monsoon storms or summer heat to crack growth on Tucson streets, so test that link against your own repair history before you build the schedule around it.

Proposition 411 directs sales tax money to neighborhood street work, so the repair list is also a funding document. A ranked list that shows which cracks are growing gives council members a record for project choices that does not depend on one inspector's memory of a street.

How do you turn drone images into a ranked repair list?

A survey is only as useful as the rule that sorts it. Before the first flight, your engineers should write the ranking inputs and their weights, so the output is a repair order rather than a gallery of photographs. The inputs that usually matter most are these.

Set aside a ground truth sample before you trust the ranking. Have experienced inspectors score a set of streets by hand, compare the results with the drone output, and tune the rule where they disagree. A ranking that agrees with your inspectors on the top of the list is more useful than a sophisticated one nobody checked.

Keep the output in the systems crews already use. The list should arrive as work orders with coordinates and photographs, so a repair crew can find the crack without a second survey and the ranking can be audited later.

• Distress type, since accuracy differs by crack type and the ranking should reflect that
• Severity and extent from the images, checked against ground samples
• Street class, so arterials and neighborhood streets are weighted for their own traffic and funding rules
• Change since the last pass, which shows which cracks are accelerating
• Existing Overall Condition Index score, so the new ranking stays comparable with past years

Are drone flights over Tucson streets legal at night and near people?

Federal rules are the starting point, and they are less restrictive than many people expect. The FAA states that Part 107 pilots can fly at night and over people without a waiver under certain conditions, a change that took effect in April 2021. Your pilot still has to meet those conditions on every flight, and the flight plan should say so.

Night details need checking before you schedule them. A secondary summary of the 2019 proposal says night flight requires anti-collision lighting visible for at least 3 statute miles and completed night training. Confirm that against the rule text in 14 CFR part 107 before you plan any survey after dark.

Part 107 includes no privacy protections, so state and local privacy rules still apply to imagery of people and property. Ask the city attorney how overflight of public right of way is handled, and keep flight logs and imagery retention rules in writing for council review.

How does crew staffing affect the survey decision?

Staffing is the other half of the backlog. The Bureau of Labor Statistics projects highway maintenance worker employment to grow about 3% from 2024 to 2034, which is slower than average, from about 159,100 jobs nationally in 2024 to about 163,900 in 2034, with about 12,300 openings per year. Those projection figures came through secondary summaries of BLS data, so confirm them on the BLS site before you cite them in a budget memo.

National figures cannot tell you how many workers Tucson can hire. What they do suggest is that slow-growing labor pools make it hard to add walking survey capacity, so the survey work is the part worth automating while crews keep their hours for repairs.

Plan the program around the people you have. A drone survey still needs a trained pilot and a reviewer who checks the output, so a small team can start with one pilot and one reviewer rather than adding field walkers.

When is a drone survey not the right fit for Tucson streets?

Drones are a poor match for parts of the network. Heavy tree canopy can hide the surface, and narrow streets with overhead lines or tight turns raise flight risk. Pavement under fresh sealant or loose debris may hide or distort cracks, and a model trained on a small image set may miss them, so those streets need a human pass.

Drones also do not measure what lies under the surface. If the question is base failure, subgrade moisture or drainage, a surface survey will not answer it, and you will need cores, test pits or a geotechnical study. The drone list can still tell you which sections to sample first.

Test before you commit. Run one neighborhood street set and one arterial, score both with your inspectors and compare the results. If the agreement holds on your own pavement, expand. If it does not, you have spent a few flights, not a fleet of aircraft.

How would Tucson start a drone pilot without a long commitment?

Start small and make the pilot prove the ranking. One district, a fixed flight schedule and a written scoring rule let you measure the drone list against your inspectors before the spring survey becomes official. Choose a district with a mix of arterial and neighborhood streets so the test covers both.

We work as an OEM-neutral robot integrator, so the aircraft and sensor setup are chosen for your streets rather than from one manufacturer's line. We lease the robot through financing, deploy and integrate it with your work order system, train your pilot and reviewers, and service each unit under our service plan. Service includes on-site dispatch, remote triage and regional service engineers, and the lease is quoted after a free site assessment that maps the pilot route and checks flight conditions.

Ask for a pilot that starts with one leased unit and a backup plan for downtime, so a grounded aircraft does not stop the spring list. Set agreed accuracy thresholds before the first flight, and decide on a district-wide rollout only after the numbers from your own streets come back.