There was once the notion that autonomy was not ready, when in fact autonomy has been viable for various missions for many years. The spectrum of autonomy covers everything from obstacle avoidance on remotely operated vessels to the full autonomy, including COLREGS behaviors, resident on Sea Hunter. There is lots of spectrum between those two extremes. I would submit that’s not where our issues have been for some time.
UNRELIABLE RADAR
Perception, or “seeing” what’s around the vessels needs to be highly reliable and effective for USVs. We can build autonomy that is safer in operation than having a complete bridge watch team, but that autonomy is only as good as the perception system that feeds its world model. For years, radar was 90% of the perception input for USVs, but radar has major weaknesses including clutter rejection, lack of nearfield coverage and, in some cases, just plain lack of resolution. High sea states and bad weather exacerbate those issues and, as a result, radar should never be considered the sole perception sensor for USVs.
CAMERA CONSTRAINTS
Humans understand “seeing” things and cameras provide remote operators a readily understandable view of the vessel’s surroundings. When quality EO/IR cameras are utilized, they can be an “acceptable” solution for many applications where a remote operator, or “man in the loop” is employed. Many USVs, including our own, now employ cameras as another part of the autonomy perception capability. Not as situational awareness for remote operators, but as input to the autonomy world model picture.
Autonomy requires more. Cameras also fail in many scenarios—heavy rain, fog, high sea states, low contrast scenes like nearshore coastlines, etc.—so they are not the final solution either (even combined with radar).
TRACKING INFORMATION
When USVs are fully autonomous and there is no intended constant oversight, the EO/IR perception must provide not only vision, but context. In that case, autonomy needs track information—contacts must be detected and tracked (and in some cases identified) in order to provide world models and path planners the vectors from which to make maneuver decisions. Track information must always be available—day, night, bad weather, etc.—if USVs are to operate 24/7/365. A single lapse in ability to detect potential hazards could result in loss of life or damage to vessels. That would be a tragedy the industry would find difficult to overcome.
Leidos and others are working with EO/IR companies, signal processors, and artificial intelligence developers to find ways to enhance optical perception for use on board USVs. That’s only part of a more complete solution that will involve multi-spectral, multi-modal sensing that provides autonomous platforms with the constant, high 99th percentile perception picture we need to ensure we continue on an unimpeded course to widespread adoption of USVs in defense and commercial applications.
This spotlight appeared in ON&T Magazine’s 2025 November Edition, Remote Operations & Force Multiplication, to read more access the magazine here.
