Autonomous with no ground station?

Please help! I’m new to this field other than doing a bit of flying with COTS things (Blade Nano CPS mostly). I’m now in a team of 5 doing a project for a final MEng module and we’re set to design an autonomous drone!
This is no ordinary drone.
It has a 100km range, weighs in at 130kg, powered by 2x 12kW electric motors. I suppose it’s mostly like a fixed wing but is more UFO shaped (but that’s another argument!).

What I’m trying to find out is how to navigate it. The plan was to stick a rugged Android phone inside and let that control everything. I’m now thinking I’ll at least need some other hardware like an APM board. But the biggest question is about the true autonomy side of things. We need to be able to select a location on the phone, stick it to the drone and launch it and let it fly there with no ground control station of any kind.

Is that a thing? Have I missed it somewhere? All my searching comes back to having a ground station and/or someone with a transmitter within range.

I’m ignoring the law at the moment as well :smiley: , we have another team member looking into regulatory stuff, I’ve just got to assume we can fly anywhere at the moment.

Have a look at this and see if it changes your mind about not having a ground station.

http://www.dragonlinkrc.com/

A ground station can be a smart phone or tablet or a collection of impressive hardware.And autonomous has several meanings.Truly autonomous and the drone would be making decisions for itself.A programmed in mission can look as though it is autonomous but is just playing a pre-recorded flight path.You have to program everything in correctly beforehand.Using a Pixhawk you can program in a flight path and set it off with instructions to ignore if it loses contact with the ground and continue the mission but, again, you have to program it right.

If it’s not packed full with military grade sensors then you are not going to get a fully autonomous flight.Programming in a long range mission and setting it off has been happening in ardupilot for years.Easily do-able so it depends on the end use of this mission as to how you go.You can have long range control and video or you can launch it and go down the pub.

http://ardupilot.org/copter/docs/common-mission-planning.html

2 Likes

Thanks! Our brief is to create “a UAV” that:
• is autonomous.
• is powered.
• can travel >100km
• can carry >50kg
and some other bits.
So I guess we can interpret that as required. The goal is for a basically trained person to be able to launch it from any site and it land where they are told it needs to land. So there’s going to have to be some level of decision making on board based on the landscape - maybe we can just tell it to climb high enough to avoid low things and low enough to avoid high things. Mountains might be an issue.
Landing without hitting people is going to be another concern, it may have to do some kind of image analysis (OpenCV?) and adjust it’s position based on that (landing is likely to be via parachute). If that’s something we’re doing we could use that for basic intelligence during the flight as well.

Yep.It’s the parameters of “autonomous” that matter.We’ve been flying missions for years with the Ardupilot code and calling them autonomous missions but they’re just preprogrammed routes.If anything gets in the way or changes then - fail.It can’t go around stuff without prior programming.There are short range sensors available that can avoid things at lowish speed and shortish range but I doubt that applies to a big bird like that.

I have no idea where you’d go for advice on a system for real time on board monitoring of the environment and the ability to react to it.There may be someone here with the brains to help or a link to the info.You’re certainly looking at co-computers (I’ve got a couple) for number crunching and decision making linked to the flight controller which would be doing the flying.That’s already common and there’s a section on here for that stuff.

https://discuss.ardupilot.org/c/apsync-companion-computers

There are also other considerations with something that large.it’s no longer a small UAV buit a large one.The rules change at 20kg.At 150kg it starts to become a real aircraft with real CAA requirements.if you’re bolting on 50Kg to the 130Kg then you hit that limit.There will be big rules for big planes.

http://www.caa.co.uk/Commercial-industry/Aircraft/Unmanned-aircraft/Large-unmanned-aircraft/

You have your work cut out so best be informed as to the intentions of the enemy.This is the LUAV pdf including all you need to know about really long pdfs (165 pages).You’ll need to know this stuff I’m afraid.I can ignore it for the time being. :grinning:

http://publicapps.caa.co.uk/docs/33/CAP%20722%20Sixth%20Edition%20March%202015.pdf

Part of this document actually helps with a definition of autonomous. -

Autonomous Operation - An operation during which an unmanned aircraft is operating
without pilot intervention in the management of flight.

It doesn’t say anything about not hitting anyone.

And also,interestingly - Autonomous Aircraft An unmanned aircraft that does not allow pilot intervention in the
management of the flight.

You’re not Amazon in disguise are you ? :joy:

We’ve purposefully stayed below the 150kg limit, the craft is the payload. And no… it’s not a military payload, it’s humanitarian (a water filtration system, power generation, communications etc.), so hitting people is something we really do want to avoid :smiley:

I’ll have a read through that stuff though, or pass it on to the legal team member to look at. For the Co-Computer would a modern mobile phone be appropriate? Or are we talking much more power? runs off to read the other forum

All the co-computers I’ve seen are linux based.So no idea if a mobile phone would work.Possibly I’d have thought.They have enough computing power certainly.

Sounds like an exciting project and I wish you luck.

A great challenge!
ADS-B highly recommended, as is something like Altitude Angel (https://www.google.com.ec/amp/s/blog.altitudeangel.com/2017/06/26/worlds-most-popular-open-source-drone-control-app-now-connected-to-global-utm-system/amp/).
One key challenge is ensuring that you maintain the ability to intervene: ie even when operating autonomously, without a human-in-the-loop, there is a human-on-the-loop who can hit a kill/pause/rtl switch

1 Like

Sir,
Can you please upload some image of your model, so we can better understand . . .

We only have an initial concept at the moment, we’re still waiting for our aero guy to show us the current design - he seems to love doing it in secret and doing a big reveal, despite us needing to see it to continue our own parts of the project.

Best of luck . . . . .

Not sure how to go about doing this since the mission is meant to be autonomous Thanks.

We’re looking to have some onboard BGAN network, so we could connect to that. I’m not sure how satphones handle high speed movement though! I guess we’re lucky that our design is 100kg and has onboard PV :smiley:
I’m fairly sure the OTS apps available will be able to fly it, my issue is not knowing much programming and needing to find a way to launch it - as we won’t have a pilot at the launch site, it’ll need to “know” when it’s being launched and activate the right apps at the right times.

One thing I’m not sure on is how to upload a mission (from Mission Planner for eg) without a transmitter attached in any way. Do you have to have a tx for initial take off etc.? Ideally we want some kind of connection from a laptop to the drone - whether we have a FlytPOD or something in there, download the mission with waypoints already set, and then launch it manually and have the missions begin after a set time.

Can you do that?

Maybe a USB connection from laptop to FlytPOD? Download mission? Some kind of mobile phone connection with an app that they can hit “GO” on? USB connection from FlytPOD to mobile?