A review of underwater robot technologies for inland waterway inspection and maintenance: the pinglu canal scenario

Underwater robotics has advanced fastest in its individual capabilities. Learned optical enhancement, sonar interpretation, and tightly coupled multimodal fusion each report steady gains on dedicated benchmarks, yet whether those gains translate into reliable operation under deployment constraints remains largely untested. We examine this question for confined inland waterways, where narrow geometry, persistent turbidity, dense engineered structures, and GNSS denial jointly decide what a robot can and cannot do. The Pinglu Canal in Guangxi, China, serves as the reference scenario, and the analysis is organized around a scenario-centric evaluation framework rather than an isolated capability survey. Across perception, multimodal fusion, SLAM, planning, communication, and platform design, we assess each technology for its functional adequacy under canal-specific conditions, then map the resulting capabilities onto bathymetric survey, infrastructure inspection, hazard detection, environmental monitoring, and lifecycle maintenance. The reviewed evidence points to one conclusion. No reviewed technology is adequate on all three axes of the scoring framework applied here, and no reviewed study demonstrates an integrated system operating under the combined canal constraints. The gap is system-level integration rather than algorithmic maturity. Each identified failure originates in a single module, escapes detection at that module’s own output, and becomes observable only after corrupting the layers above it. Reliable operation depends on adaptive multimodal perception paired with structure-relative localization and persistent mapping, a requirement that transfers directly to ports, bridges, tunnels, and other infrastructure-dense waterways.