I’m wondering what it means to sell an “engineered” product with regard to autonomous vehicles?
Are there actually any standards which can be applied to a usv for example for an engineer to make sure it complies with?
I’m not a degree qualified engineer but I’ve built a (well many) working platforms which satisfy fhe functional requirements of a task but i imagine many potential users of such devices wouldn’t purchase something that wasn’t “engineered”
What does it really mean in the ardupilot space and how would i go about getting my vehicle “engineered” if there is even such a thing?

That usually means writing requirements and testing that the requirements are meet. And of course documenting the entire process

So just acceptance testing like i would do for commissioning? Proving failsafes etc?

I was thinking it woild have been about calculations of structural integrity and ensuring that it was built in a way that met standards on material thickness and strength etc, but the testing that it meets requirements is far easier.

At the end of the day it needs to be compliant with applicable laws, and that varies from country to country. So do not take my word, inform yourself.

The question shouldn’t be “how do I engineer,” but rather “what requirements am I trying to meet?”

Then you can tackle the analyses and formulate tests/simulations to provide documentation that the requirements are met.

So yuri, would these requirements just be the customers specific requirements of features the vehicle would have or would they be national/international standards for autonomous vehicle design?

All too vague to answer. Maybe both. Depends on the use case (and the customer).

I’m not sure there are any one-size-fits-all standards for autonomous vehicles. And you may have to comply with other standards like emissions/environmental regulations, again highly dependent on use case.

1 Like

The vehicle is a vessel (rover) and i was thinking that an rc boat which would be on the market would not have any documentation, nor would it necessarily meet any kind of standards other than maybe some limited safety features,
The customer in this case wouldn’t know what they want other than that it was all “engineered” and compliant with all applicable rules and regulations (government organisations)

Its also tricky for me to work out what requirements i need to meet when i am the instigator of the design and there was no “scope” put to me in the first place by the potential customer, rather I was looking for a genuine use case for autonomous vehicles and found one so i proceeded to design and build one to demonstrate what can be done today with ardupilot.

You’re in the US Yuri?

I completed a prototype and have it working and tested, but now I’m still up against the engineered requirement and the percieved conflict of interest if i were to supply to my employer, other industry players would be fine as i dont work for them.

Where did this requirement come from? I “Engineer” things everyday in the day job. Sometimes they need to meet a standard (AS9100, ISO, etc), other customers ask for support data (FEA usually for what I do). Sometimes testing results to some standard either recognized (ASTM, others) or supplied is required. As Yuri said “Engineered” is pretty vague.

It was just some initial feedback from a coworker who warned me of his past experience with a cable drum roller he had made in the past which was condemned by someone else in the business because it was not engineered and had simply been fabricated at work. So he had to go and buy one from an electrical wholesaler and cut up the one which had been made in-house

Yes, US-based.

That’s a strange anecdote. What was the actual concern? Performance? Longevity? Parts availability? Liability?

If it was liability? Liable for what?

These are among the subset of questions you must answer before “engineering.” You can’t really generically engineer something. You engineer toward a goal/requirement or collection thereof.


What you’re saying makes total sense in the real world, but the way some gov organisations work in Australia is far from real world and they mostly dont know enough to know what they want, except to say that they don’t want to be blamed for anything down the road so they waste seemingly endless dollars on making sure they never make a decision themselves so they can never have to answer for anything.

On a more practical note though, it seems the drone industry could make whatever craft/vehicles/vessels were required by the market, if only the people who know ardupilot knew what the customers needed, and the industry doesn’t know what is available or possible so they don’t always consider approaching anyone to see what can be done,

I work for a bulk water industry which regularly collects algal water samples ( 10" below the surface) for analysis. They currently do this task using 2 persons on a twin engine maritime certified commercial vessel to fill an 8 oz bottle,

I already have a 20lb rover which will do this autonomously from the shore with one person using ardupilot. That was the easy bit, working out how to get my sampler into the industry is much harder.
From what i have seen on the net, currently there are few alternative unmanned options and i have evaluated the other methods used and believe mine is the simplest for this specific task and the most suited for our specific application.

I will go down the path of listing, meeting and testing the sampler meets the requirements that i believe this sampler would need to meet, the true requirements would depend on the particular customer so i cant really predict that. Then offer it as a free trial to any potential industry players who currently do manned sample collection i guess.

1 Like

With surface vehicles as long as your not driving them on public roads there are no legal requirements and there aren’t really any standards or rules like you find with aircraft. But there will be regulations out there for health and safety around industrial machinery so you cannot have spinning metal without a guard on it, needs a big red stop button. you will also need FCC ids and certificates for all your radio equipment so you cant use diy radios or anything like that. here in the UK anything that plugs into the mains will need certified as tested for electrical safety before it will be allowed to be plugged into any facility so all your chargers would probably need to be certified.

Test everything repeatedly, Any question you’re asked you want to be able to answer by saying that it was tested for that issue and this is what happened and if that fails this is the procedure and if that fails this is the backup. like what happens if something gets snagged in the propellers, battery fails, loss of rc, and have procedures for each thing. if your extra paranoid you could also have redundancy so have dual batteries, dual motors with rudders so you can still steer if one motor gets fouled. dual rc systems, dual GPS, other options could be adding a bilge pump, and making it self righting.

Have multiple independent kill switches, You want to be able to stop it with the RC remote, with an independent wireless kill switch and some big red stop buttons.

1 Like

Thanks for the reply, its all food for thought,
I have so far gone with air props to eliminate all the snagging, weed potential etc, it will even slide up onto the bank like a hovercraft.
But then theres the spinning prop hazard.

As for the emergency stop safety situation, I’m in two minds, if you have a system where the rover stops as soon as you aren’t maintaining the rc link or an independent safety link, then you lose the rtl function, if you have rtl functionality then you also have a rover driving around without the operator having any way of stopping it in an emergency situation.

Thats why i said have multiple failsafes. Have it RTL when loss of RC but have a separate wireless killswitch incase RTL goes wrong that way your still in control something like a 433mhz remote relay.