AI Drones Find Underwater Bombs Accurately in Florida

New AI drones can find underwater bombs with 100% accuracy in tests. This is a big step to make our waters safer from old dangers.
By Newsroom | Technology, Environment |

AI-POWERED DRONES EXHIBIT HIGH ACCURACY IN IDENTIFYING UNDERWATER MUNITIONS

New airborne systems, employing a blend of advanced imaging and artificial intelligence, have demonstrated a marked ability to pinpoint underwater munitions in shallow, complex marine environments. The technology, detailed in recent reports, integrates NASA-developed underwater imaging with machine learning algorithms.

This fusion of aerial sensing and AI achieves significant detection accuracy while notably minimizing false positives. Both active MiDAR sensing and multispectral passive imaging approaches, when combined with machine learning, showed strong performance in identifying targets.

Nuances in Detection Technology

Recent tests conducted at the Broad Key Research Station highlighted the capabilities of these systems. All unexploded ordnance (UXO) targets were successfully identified within the Fluid Lensing RGB imagery. The precision of these detections was quantified, with all locations and identified targets marked.

  • Active MiDAR sensing reportedly yielded the highest precision in these trials.

  • Both active and passive sensing methods, however, achieved robust detection rates.

  • The integration of machine learning appears crucial for reducing instances of misidentification.

Performance Against Challenges

Further analysis reveals the system's resilience in challenging conditions. In one instance, a model, after approximately 200 training cycles, correctly identified all 14 planted munitions within both passive and active datasets. This occurred even as the submerged objects exhibited altered color and shape due to natural biofouling.

Read More: Consulting Firms Hire AI Experts in Houston and Dearborn

  • The targets varied in size, ranging from 25 cm to 66 cm in length.

  • They were situated on a seabed characterized by natural formations like rocks and coral rubble, alongside man-made debris.

  • The active MiDAR system, in particular, proved effective in distinguishing real targets from decoy clutter, a common tactic in mine-countermeasures operations.

Technological Underpinnings

The research involved various configurations of Fluid Lensing and MiDAR technology, utilizing different spectral bands. These included:

  • 3-band (RGB) Fluid Lensing

  • 8-band (375–675 nm) MiDAR Fluid Lensing

  • 10-band (444–842 nm) Multispectral Fluid Lensing

This work represents what appears to be the first demonstration of airborne multispectral and active Fluid Lensing detecting small, biofouled ordnance in cluttered, shallow waters. The findings were published in Frontiers in Marine Science.

Frequently Asked Questions

Q: What new technology can find underwater bombs?
New airborne systems use AI and special cameras to find underwater bombs. They are very good at finding them and not mistaking other things for bombs.
Q: Where were these new AI drones tested?
The new AI drones were tested at the Broad Key Research Station in Florida.
Q: How accurate are these new AI drones at finding bombs?
In tests, the AI drones found all 14 planted bombs. They were accurate even when the bombs were old, covered in sea life, or hidden in messy areas.
Q: What makes this new technology special?
This technology is special because it uses AI with cameras that can see through water. It can find small, old bombs that are hard to see, even in shallow, cluttered water.
Q: Why is finding underwater bombs important?
Finding underwater bombs is important to make the sea safer for people and sea life. It helps remove dangerous old weapons that could harm someone.