Maritime Autonomy / Undisclosed / 2025
Retrofit autonomy for existing crewed vessels
Design and prototype of an autonomy package fitted to vessels already in service — custom interface hardware, edge compute, protocol bridging and a communications path that degrades predictably offshore.
Disciplines
- PCB design
- Edge compute
- Marine protocols
- Communications
Context
Autonomy demonstrated on a purpose-built hull proves less than it appears to. The commercially significant problem is the fleet that already exists: vessels in service, with helm and propulsion systems from a range of manufacturers and vintages, that cannot be taken out of the water for a rebuild and must remain operable by their crew afterwards.
Retrofit turns an autonomy problem into an integration problem. The autonomy stack is the visible part; the work is in the interface to equipment that was never specified with an external controller in mind, and in behaving safely when that equipment responds in ways its documentation does not describe.
Approach
We designed the interface hardware rather than adapting a development board, because the electrical and environmental conditions on a working vessel — supply quality, vibration, salt, temperature range — are where an adapted prototype fails, and it fails at sea rather than on a bench. The board bridges the vessel’s existing marine protocols to the autonomy stack and carries the edge compute that runs it.
Communications were treated as a property that varies rather than a service that exists. Offshore, link availability changes with weather, distance and traffic, so the system’s behaviour is defined for each state of the link, including its total absence. The corresponding rule for handover is that the vessel returns to a safe, predictable state and gives control back to the crew rather than continuing on a stale picture — a decision that constrained the architecture from the outset rather than being added as a safety layer at the end.
Outcome
A working prototype demonstrating the retrofit path on vessels in service, with the interface hardware, protocol bridge and comms behaviour proven together rather than in isolation. The design is documented for manufacture, and the constraints that shaped it are recorded, which matters most for the equipment interfaces — the part a future integrator will otherwise rediscover the hard way.