Sonar-vision fusion helps underwater robots navigate cloudy sediment
A new system combines sonar and visual data to create real-time 3D maps for remotely operated underwater vehicles. It allows vehicles to see through sediment clouds and safely approach objects for detailed inspection. Applications include scientific exploration, underwater construction, and handling undersea mines.
The system pairs sonar's all-weather reliability with a depth-estimation algorithm developed in France that rapidly calculates pixel-level depth from 2D imagery. This fusion lets the vehicle build real-time 3D maps without waiting for sediment to clear, a process that previously stalled operations for extended periods.
In tank tests, the technology mapped the full environment while resolving objects at centimeter scale even under cloudy conditions. The researchers envision uses ranging from deep-sea science to infrastructure maintenance and safely approaching unexploded ordnance, where visibility is often severely compromised.
This technology could significantly reduce downtime and risk in underwater operations, benefiting industries like offshore energy, marine construction, and defense. For scientists, it may enable safer exploration of fragile ecosystems without disturbing them. Mine-clearance crews could approach hazards with greater confidence, potentially saving lives. However, improved underwater navigation also raises questions about undersea surveillance capabilities, which may warrant careful consideration by regulators and the public.