Two announcements, different jobs
Anduril’s March 12 company release, reproduced by Naval News, announced selection for the Defense Innovation Unit and Navy Combat Autonomous Maritime Platform project. The company described a planned long-duration, operationally representative Dive-XL demonstration within four months of award. Anduril also reported cumulative experience across its autonomous undersea vehicles of more than 42,355 kilometers and 6,752 mission hours. Those company-reported fleet-wide totals provide development context; they are not the record of a single Dive-XL or an operational certification. This was a prototype and experimentation step toward wider use, rather than a declaration that the Navy had fielded a complete operational fleet.
On March 25, L3Harris announced an other transaction award to provide its Torpedo Tube Launch and Recovery system for Iver4 900 autonomous underwater vehicles. The company said U.S. and allied navies had validated the system for intelligence, surveillance, reconnaissance, mine detection, and seabed missions. Those are supplier statements about the system, not evidence of installation on every submarine class.
The announcements occurred in the same month, almost two weeks apart. They illustrate different pieces of an undersea force rather than proving a single coordinated acquisition decision.
Dive-XL emphasizes the payload and endurance possibilities of a larger independent platform. Torpedo-tube launch and recovery addresses how a smaller vehicle can extend a submarine’s reach without bringing people into the vehicle’s operating area. Neither approach eliminates its own integration, support, or recovery requirements.
Deployment and recovery deserve equal attention
An autonomous vehicle that completes a useful mission but cannot be recovered predictably can impose a significant burden on the supported force. The launch method, handling equipment, storage, batteries, payload configuration, and recovery procedure all belong in the system design.
Torpedo-tube compatibility is particularly consequential because a vehicle must work within the host platform’s physical and safety constraints. A vendor’s successful demonstration does not establish immediate, modification-free compatibility with every Virginia- or Los Angeles-class submarine. Fleet integration requires the applicable platform-specific evidence and approvals.
Large independent vehicles face another set of questions: transport to the operating area, launch facilities, maintenance access, energy replenishment, and recovery after a fault. These considerations influence how much undersea presence a fleet can sustain, even when vehicle endurance is strong.
Autonomy below the communications layer
Underwater vehicles cannot assume the continuous, high-bandwidth radio links available to many terrestrial systems. Acoustic links have constraints in bandwidth, latency, and propagation; surfacing or using a communications mast changes the operating conditions and can affect exposure. The mission therefore needs an explicit plan for the periods when a command link is unavailable.
Onboard functions may include navigation, sensor processing, obstacle handling, energy management, and execution of a bounded mission plan. The system should recognize when uncertainty or equipment health makes continued operation inappropriate, then follow a tested contingency.
That capability does not necessarily require a large generative model or unrestricted learning during the mission. Conventional control, planning, and perception methods may be appropriate for many functions. Where machine learning is used, the team must account for its data dependencies, failure modes, and computational cost.
Compute is part of the vehicle trade space. Additional processing consumes power, generates heat, and may reduce endurance or payload capacity. Model compression and hardware acceleration can help, but their effect on mission performance must be measured on the actual deployment configuration.
A useful autonomy acceptance plan
For an undersea program, the following sequence makes technical claims testable:
- Define the mission envelope. State the environmental, navigation, endurance, and payload conditions covered by the evaluation.
- Bound the authority. Specify which adaptations the vehicle may make independently and which require communication with a human authority.
- Test degraded information. Include missing or ambiguous sensor inputs, navigation uncertainty, and interruptions to external assistance.
- Prove recovery behavior. Exercise equipment faults, energy limits, interrupted missions, and loss of the expected recovery opportunity.
- Control software and payload changes. Record configurations and identify what must be retested after an update.
- Inspect the mission record. Preserve enough telemetry and event history to distinguish a successful mission from an unexplained outcome.
These are recommendations for engineering assurance, not a claim that either public award announcement specifies this exact checklist.
Scale depends on interfaces and support
The acquisition route can help a nontraditional supplier reach a prototype opportunity. It does not remove the need for a production decision, an executable support plan, and integration with the supported force.
The department’s Modular Open Systems Approach is relevant because vehicles, payloads, mission software, and command systems will evolve at different rates. Verifiable interfaces and usable technical rights can reduce the cost of replacing a component or introducing a new supplier. Security controls must extend to mission loading, software updates, stored information, and authorized access to connected systems.
At fleet scale, coordination also needs realistic assumptions about communication. Vehicles should be useful within their assigned authority when isolated, while sharing information and reconciling state when a link becomes available. A safe contingency may be the correct outcome for a damaged or uncertain vehicle; operational usefulness includes knowing when to stop.
The March announcements show why hull design and autonomy architecture must advance together. The enduring opportunity is a supportable undersea capability whose behavior, integration limits, and recovery plan are understood well enough for a commander to use it with confidence.
Sources and further reading
- Anduril CAMP announcement, company release reproduced by Naval News
- L3Harris: Torpedo Tube Launch and Recovery award
- Departmental Modular Open Systems Approach guidance
Spartan X’s autonomy, edge-computing, and systems engineering work addresses the decisions beneath these platforms: what runs onboard, how behavior is verified, and how the system continues safely when connectivity disappears. Those choices shape the operational value of the vehicle as much as its physical design.
