Can I make a Career out of this?

Hi all,

I’m a retired combat systems engineer, ex Canadian navy. I can say I know some about underwater systems, but I can’t say I know anything about ROVs.

I would like to get involved in the Canadian government spending boom for what is called dual use products. For example I start a business and can produce something that can benefit both the civilian and military community.

I am knowledgeable in aspects of AI. I would like to start my own work on autonomous missions in the underwater environment. It will be quite a leap for me to just buy a ROV and start driving her around, not to mention auxiliary sensors and equipment I will want. But I’m hoping to have a good enough business proposal for government funding of some kind. I’m also keenly open to research and development funding.

I guess my discussion topic is: if you think about autonomous fault finding for consumers, that’s simple enough to transfer directly to military purposes right? so what do you think of me spending my free time (I live along the ocean) on my boat for the first few months just driving and learning an ROV? And then coming home each day and messing with software and hardware. To me it sounds awesome. Can I do it?

1 Like

Hi @BuzzyQ, welcome to the forum :slight_smile:

That’s highly dependent on how it’s implemented, what kind(s) of faults it’s targeted at, and how it’s intended to be used.

As a trivial example, if the vehicles being used by the military and civilians are quite different then training a model to detect faults on one of them may not readily translate to the other, especially when it comes to differing dynamics, data streams, and integrations with things like control station software. The converse is also true - if they’re using the same vehicles and software then transferring could be very straightforward.

As a more nuanced consideration, if you develop some kind of module that gets connected to various data streams and trains itself per vehicle (or fleet) to identify when things seem abnormal, that may be readily transferable technologically, but may have stricter regulatory requirements and guarantees before it’s allowed to be used on military vehicles. If that kind of model doesn’t need to know what kind of data it’s processing then that would likely make it more broadly usable and easier to pass at least some acceptance requirements, but would then also likely require a more extensive training period at the start of each integration where it learns what “normal” looks like.

Sounds like lots of fun! Whether it’s conducive to starting a viable business depends a lot on you and the circumstances/environment you’re in, which I at least can’t meaningfully comment on or predict.


For some more general commentary, I do think fault detection is an area of underwater robotics that is likely to be quite impactful, and isn’t meaningfully widespread yet, but exactly what form(s) the successful implementations will take is dependent on market needs and forces that many of us are likely not well-versed in.

As a few off the cuff potential options, there could be live models that run in real time either on the vehicle (for self identification) and/or on the control station software (for supervisory identification to a pilot/operator), and there could also be cloud based services that do bulk, batch processing of logs to help identify potential issues across a fleet of vehicles (possibly across multiple organisations), and are used to tune maintenance and inspection schedules and the like. Live cloud-based options could also be possible, but seem a bit less likely to catch on because they require vehicles or operators to consistently be connected to the internet, which often isn’t practical (and opens some additional security and data protection risks).

1 Like

Nice to meet you @EliotBR

While the logistics of a viable business are up in the air, there’s still a lot more I need to learn as well.

I’ve been reading papers. But of course that’s all hypothetical to me, until I can start to practice. For example a lot of papers discuss AI computer and decision making happening on a surface vessel. I would like to spend my time figuring out: what is the best AI setup on board for the sensors I have. It’s interesting to think about AI also controller a sonar for example (parameters), for it’s best use in each moment of time.

I’m all ears if you have any good reads to suggest to me :slight_smile:

As I apply for research and grant money, do you think I could viably be a pilot after a couple of months of practice of I had my own rig? The government will want me to supplement my own income as well, so I’m thinking I could maybe do local “easy” hires on the side.

Thanks very much for the discussion!

Edit: I forgot to acknowledge your point about the military/civilian dual use. I think once I’m further down the road, my company would have to split between the civilian designs and military design. Which will use similar concepts I would have to imagine, but are designed differently. What will my machine offer that others done? Maybe nothing, maybe I’ll just do it better and faster? I’m not sure but that’s where the government money will have to come in to fund me while not alot of business is happening.

Hi @BuzzyQ - you may find some of the work my colleague Isaac did recently for the ROS maritime working group & Open Ocean Software useful:

2 Likes

I might be able to help. I’m currently working on a mini subsea glider, and am familiar with the Ocean Tech start up space in Nova Scotia. I’ve found there is lots of dual use finding out there, but it seems to be targeted at companies that are already profitable through their own sales channels. But there are still lots of other grants out there for us little guys.

If your interested I could set up a teams call and talk about my experience down this path so far.

1 Like

Thank you for this! I will be reading through this right after this message!

Thank you for this! I will be reading through right after this reply to you posts :slight_smile: