Non-Invasive Inspection of Water Canals Using Dronar
This arXiv preprint presents dronar, a custom unmanned surface vehicle that carries a consumer Lowrance sonar unit to inspect concrete water canals without draining them. In Arizona Canal field tests, the system detected sediment accumulation as depth offsets up to 15 cm versus a cleaned segment, with repeated clean runs agreeing within 7 cm. It offers utilities a lower-cost way to prioritize canal maintenance between the current four-year dry-up cycles.
Can a low-cost, off-the-shelf sonar and USV system non-invasively detect sediment buildup and produce repeatable canal-bed depth measurements in operational concrete canals?
Open concrete canals accumulate sediment and develop cracks or lining deformation; current inspection relies on visual checks above water and draining whole segments every four years, which is costly, infrequent, and cannot prioritize underwater defects.
Prior sonar mapping required specialized vessels and trained operators; more recent work mounted consumer fish-finder sonar on aerial drones or small watercraft. Aerial drone-tethered sonar suffers from airspace restrictions near airports, trajectory and positioning mismatches in narrow canals, and short flight endurance. Salt River Project's current visual plus dry-up practice misses submerged and sediment-covered defects.
The authors built a twin-pontoon USV with differential electric thrusters, CubePilot Cube Orange+ running ArduPilot rover firmware, Herelink telemetry, and a Lowrance Active Imaging HD 3-in-1 transducer mounted under the waterline. It operates untethered in Manual or Auto mode, uses a custom cart for launch and recovery on steep canal walls, and collects DownScan and SideScan logs. Depth profiles are band-pass filtered to separate local bed features from background canal grade.
Over three field tests, the 16 kg USV completed a 1.6 km downstream survey in about 25 minutes at 8 km/h and used 38% battery; upstream speed was only 1.6 km/h, so later tests used downstream-only passes. A roughly 400 m sedimented segment was surveyed before and after cleaning: pre-maintenance DownScan depths were approximately 15 cm shallower than post-maintenance, and filtered pre-maintenance residuals reached about -15 cm while post-maintenance residuals stayed within about 5 cm of zero. Across six clean runs, repeated depth profiles matched within a maximum of 7 cm, with total measured bed depth range around 12 cm. SideScan data were collected but not quantitatively analyzed.
The authors note validation is limited to sedimented-versus-clean comparisons and only approximates deposit size and depth; they did not test cracks, liner deformation, foreign objects, or sediment types. Processing was manual Python scripting, SideScan was unused for analysis, and shallow DownScan coverage leaves canal margins and wall-adjacent areas unmapped. The study is a preprint with three field tests in one canal system, downstream-only operation, and one custom USV. The Herelink controller overheated at about 35°C, and recovery required practice, with the first recovery needing four attempts.
Water utilities, irrigation districts, and canal operators such as Salt River Project could use this for routine condition surveys and maintenance prioritization. Inspection service providers or drone/USV survey firms could offer it as a service. Near term, pilot deployments and manual data review are possible. Wider operational use in 1-3 years would require automated processing, validation across defect types, SideScan integration, and battery and thermal management for long canal networks.
The full text is not republished here because the paper's license does not allow it. Read the original on arXiv.