Open Pilot : new open source multirotor platform

bob.titus

February 24, 2010

this seems quite interesting :

http://www.openpilot.org

Can't wait to see what will come out of this new project.

Bob

03:34 AM

dankers

February 26, 2010

Thanks for this thread, been meaning to post about OpenPilot for ages.

I can explain a lot more and I hope people will really love what we are trying to achieve.

This project is fairly unique because of the way it is being run. We have put so much thought in to doing things the right way to start with and of course there is no point doing a project the same as everyone else!

OpenPilot stands on the shoulders of giants for sure, in some cases we have seen what they do right and used that, in other cases we have seen what other projects have done badly and gone in a different direction to avoid it. It is a unique project and I am extremely excited about it.

The key differences:

Non-Commercial project. Money ruins things at times and this is meant to be fun, hence we took money out of it as much as possible. We want the project to be accessible to everyone and the aim is to build a strong community. If people want to build a business around OpenPilot, they are free to do that and that would be excellent from the projects point of view. However, the project its self is set up for fun and not profit.

Obviously being a community project we are fully Open Source, in fact one better, we are what is called Free software and in both senses of the word: Free as in Freedom and Free as in we do not charge for it.

Hardware. Simply, it is really is awesome. Hopefully some people with hardware experience will chime in and confirm this rather than taking my word for it. Being non-commercial we can make awesome hardware and sell it for a very very good price. This is really a game changer and nothing like this has been tried before.

The project makes the hardware. Again a key difference, other Open Source project put everything out there and use vendors to make the hardware for profit. We don't. We are getting the hardware made and selling it as cheap as we can. This is about building a community and something cool. I have already put in a lot of my own money in to this so my money is truly where my mouth is. We also want to control the quality and everything is from a RoHS certified PCB assembler.

Fully finished hardware. A small but important one, the hardware is ready to go when it is sold, no soldering left to do. We are doing a lot of testing to make sure when the hardware is avilible, it will be right first time.

Built from the ground up as a UAV. OpenPilot from day one has been built to have advanced features like position hold and return to home. Additionally there are no limits to distance or waypoints, of course being Open Source there would be no point doing this anyway as it is so simple to undo.

We have used the industry standard way of doing things by using an AHRS. This is another major advantage we have and another example of things done right. The AHRS contains most of the flight sensors as well as a very powerful 32 bit processor. The AHRS handles all the complex maths of attitude estimation and this frees up the main processor for other tasks. This is a high end solution and standard in the UAV industry.

Besides these there is a lot more going on as well, for example anyone flying Quads (and especially Octos) should be interested in our TurboPWM [ http://www.openpilot.org/TurboPWM ] feature.

This is just one example of where we're rethinking everything, where a suitable solution exists we use it, where we think we can do better, we do.

A lot of the time has gone in to the Wiki [ http://www.openpilot.org ], I think we are 10 weeks in to this now and we have a very impressive amount of work done. We are also trying to make it simple for new developers and naturally a lot of work has gone in to this first. However, we have a few cool things for users, for example a lot of new people want simple and cheap frames, looking around we found not that much that was easy and cheap for a newbie so we got some made, again they are sold as cheap as possible:

http://www.openpilot.org/OpenPilot_Easy_Quad

The pricing is on http://store.openpilot.org for everything. It is not open yet but the prices are real. We have been spending so long developing we only just got around to the process of opening a bank account, when that is done the store will be open.

Just please respect what we are doing, we are a Community project, by the community and for the community. If you can, please get involved.

Despite all the work behind the scenes, we only announced the project in the last couple of weeks on the DiyDrones PodCast [ http://diydrones.com/profiles/blogs/podcast-episode-16-david-1 ].

The MP3 can be downloaded here [ http://diydrones.s3.amazonaws.com/mp3/diydrones%20-%20episode%2016.mp3 ].

It would be worth listening to that to get some more information regarding OpenPilot [ http://www.openpilot.org ].

I hope you like what we are doing, we have forums at forums.openpilot.org, come along and say heelo and catch up with how we are going. Of course if you can help in anyway, that would be really appreciated as well.

04:57 AM

schooner2000

February 26, 2010

dankers - great post, thanks for the info.

I'm just a quad noob at this point doing a lot of reading but this project really interests me. I'm hoping to build my first quad sometime this year. What sort of timeline are you looking ta for having a fully functional quad board(s) ready for release and real use? 1 month, 6 months, year? Just trying to get a sense of where things are at overall.

Also, will you be putting together a "suggested" full parts list for the new quad builder, that includes everything to put together a workign system under the OpenPilot platform?

05:19 AM

dankers

February 26, 2010

It is very very hard to answer the time question. Within 6 months for sure but longer than 1 month. It is so tempting to say 3 months but that is a kind of internal time frame, with projects like this, things can change. We have a very talented group of people, our hardware guy is simply amazing and we are working very very hard on things. The more help we get the faster thing will go.

Hardware wise, the main OpenPilot board is done, several developers have them, as always there could be some small issues but they will get resolved quickly. The power sensor board is done (optional part), the AHRS will be completed next week and the GPS boards are designed, just need to order them. We are also testing two GPSs, both 10Hz and both seem to be very good but we want the best so we are still evaluating them.

Basically we can say the hardware is almost complete and a lot of work went in to it, hours and hours of testing. That is a big milestone and done in record time, Angus who did the hardware had many weeks with no or little sleep but the results are really amazing and everyone is real proud of it.

So, that leaves us with software, the foundations are in place but this is where most the work remains. The nice thing about a project like OpenPilot is that it is modular so tasks can be split up but I won;t pretend we are close to finishing it, we are not. There is loads to do still.

There are progress updates on the forums and some developers have still not got their hardware yet, when they do things will step up a gear!

06:10 AM

dankers

February 26, 2010

Quote:
Originally Posted by schooner2000
Also, will you be putting together a "suggested" full parts list for the new quad builder, that includes everything to put together a workign system under the OpenPilot platform?

We will go much further than this. We have Juan managing the "getting started" page on the Wiki and I know that will be exactly what you need. But there is more, there will also be a tutorial video that shows starting from scratch to a flying Quad. Other tutorial videos are also planned, such as flying your first waypoint, updating the firmware etc etc. For example see the frame video on the Easy Quad Page:

http://wiki.openpilot.org/OpenPilot_Easy_Quad

For developers new to the STM32 we did this:

http://wiki.openpilot.org/Eclipse_Setup#Tutorial_Video

It is exactly the kind of thing we will continue to do as it makes everything easy.

I am still working on that Easy Quad page as there is a very simple and effect Carbon Fiber upgrade that we came up with. Again being non-commercial we have the advantage of just showing these things for everyone to benefit from as we can't lose any money by doing so.

For people flying Quads now they see it as second nature, when I started I came from a fairly experienced heli background but was lost with quads. I have made a load of notes on where I was stuck at the time and we will make sure everything is covered. Thanks to the people on RCGroups I got in the air and learnt a lot quickly but when I started, wow, what boards do I need? Which versions? What needs Soldering? We are trying to do as much as we can to stop this as well. Like I said everything is finished, no soldering. Another big one, the USB connector is on the board and it speaks via HID to the PC, no drivers required on Windows, Mac or Linux. No sercon or any or USB to serial, just plug it in and that's it. A very nice feature from Angus and the code for this is already done.

So yeap, there will be a kit on the store with everything you can buy from us. The other place we will show links to is Hobbyking, no point us trying to stock motors and all that and they are very cheap.

We might stock props if Hobbyking don't in the time we are ready for users, simply because I know how frustrating and expensive the shipping can if you have to hunt down all the different parts from different stores. Hence we are trying to keep it to two and as HobbyKing has good prices and the place we use ourselves.

06:48 AM

ddmt

March 01, 2010

Just wondering, how long would it take for openpilot to be the same level (as in feature like position hold, etc) as the Mikrokopter?

08:02 AM

schooner2000

March 01, 2010

I think they are targetting 3-6 months to be at that level.

08:05 AM

dankers

March 01, 2010

Exactly, around 3 to 6 months for that level. I think considering that it took MikroKopter a year to get their first Flight Control done we are doing very well as ours was done in under 4 weeks. I think MK started in 2006 and had something out by 2008 but that had to be all soldered by hand with no ready build hardware.

The biggest goal right now is getting in the air, it's all hands on deck for this. Once this is achieved I hope we will complete the UAV functions like position hold and RTH at a faster rate. The reason why I say this is because the code right now is a lot closer to the hardware, it is still laying the solid foundations we need. Actually a bit of news, last night thanks to the work done by some talented people we were running up motors connected to a MikroKopter I2C ESC. So we now have I2C Esc support working, it just needs tidying up a bit.

One of the big heroes in this was Rowan who found us thanks to this thread. He figured out our hardware issue that was blocking this and got it fixed in a couple of hours. I actually can't think of a better example of how the community is meant to work.

This a huge difference between us and MikroKopter, they have a community and have some genius people in it, however the project can not easily adopt what they do but because their commercial nature. This is why there are so many branches of MikroKopter code, for example ArthurP did some nice changes to the altitude hold code but this can not be adopted back in to the project because the owners worry about licensing. We do not have that issue and if people find issues or fixes we welcome that with open arms.

We are very democratic and when some very smart people work on things together, the results can be fantastic. Contrast this with MikroKopter which is run more along the lines of a dictatorship, not that the owners are not nice people but, just in the sense of what they say goes.

Saying all this MikroKopter are not a project we set our sights on, it is just a quad controller with some UAV features. Our architecture is very different from MikroKopter's, our hardware just blows theirs completely out of the water, they really are not the kind of level we are aiming for.

MikroKopter is aimed at Quads and Hexa etc, we are starting with these as they enable us to get in the air quickly and are an awesome platform but if done right a real UAV should be able to fly Helicopters and fixed wing aircraft as well. MikroKopter lack of advanced filtering and the overall platform design means they are really a multi-rotor platform and basically very much like the AeroQuad project with some UAV features tacked on.

The idea with OpenPilot is to take on MicroPilot [ http://www.micropilot.com/products-mp2028-autopilots.htm ] and projects at that level, these high end commercial projects are very good and this level of quality is where the challenge is.

So, for people looking for a Quad platform, you are going to get something that is a really awesome, something that is very high end. Our hardware is designed to the level of MicroPilot and surpasses it in many areas.

One of the things I really urge against is to judge OpenPilot by its price, if you listen to the podcast I explain all this there. The OpenPilot board, AHRS board and the GPS will sell for less than just the MikroKopter ME board. The reason behind this is because we want to get our "toys" out there in the real world doing some good. There are many civilian applications of UAVs that can help save lives and that is the idea. If you want to use a UAV for looking for lost climbers or looking for earthquake survivors etc you need something much more serious than MikroKopter.

I hate vapour ware so I'm not going to go too much in to features etc but I hope that explains a bit more about the goals and ideas behind OpenPilot.

Small UAV news [ http://www.suasnews.com/2010/02/241/new-open-source-uav-autopilot-launched/ ] called us a potential game changer, that really is the hope of the project.

Finally, don't be put off by all this talk of UAVs, if all you want a very good and well priced Quad / Hexa / Octo platform not only can OpenPilot do it (MicroPilot actually can't) but it is what we are building our foundations on and where we need the most people to help.

01:35 PM

bob.titus

March 25, 2010

Good point Dankers

@ Anthony : good ideas will always be copied but today they can be copied before you even have a chance to use them yourself, I find this very annoying when you know the time spent in development to get there.

Bob

06:35 PM

dankers

March 26, 2010

Quote:
Originally Posted by bob.titus
@ Anthony : good ideas will always be copied but today they can be copied before you even have a chance to use them yourself, I find this very annoying when you know the time spent in development to get there.
Bob

Bob is right and knows some of the history of what happened to us & this is why Angus is not saying everything about RTH, if we did it would likely appear on someone else's web site as a planned feature and "their" innovation when really it is just taken from this thread. The hardware has no profit in it, the software is out there for download by anyone, all we ask is at least give the project the credit for our new features. It is also a community project so we were hopeing people would help, it is a project for everyone.

So much time and effort has gone in to the planning of this project and we have ideas we have seen nowhere else. So, against my better judgement here's the RTH stuff based on long discussions, it shows how our architecture design enables us to do some very cool thing.

There will be three RTH options to start with:

1. Direct
2. Safety Altitude
3. Retrace path

Direct is simple, come home in a direct path, hit a tree, opps. Same thing as MK does. Safety altitude is direct but will climb to a set safety height before returning home, you can optionally descend when home is reached.

Retrace path will run the way points in reverse to reach home and this is something I have no seen done anywhere before. So to start with, this will be a UAV feature and work only when a flight plan is used, it will retrace basically the flight plan.

The version after this will get a bit smarter and log the whole flight to the SDCard, it with automatically generate return waypoints on the fly, this way the retrace feature will work even in free flight mode. This is cool but one thing to watch is battery life, so in this mode at the end of a long flight, you will not want to retrace everything. OpenPilot is script-able so can be told to switch to direct mode when a predetermined battery level is reached. Like a lot of things, the operator has to select which is right for them.

I hope this explains something no one picked up on, in a earlier post I said our three positions switch would be:

1. Normal free flight
2. Return to Launch
3. Autonomous mode.

The retrace path feature is why we went this way, for people that know MK they use the RTH as also the Autonomous mode, we want to have a RTH while in autonomous mode hence the different way this is done in OpenPilot.

All this is based on the "Record and Replay" module (or as we call it R&R), out of this comes some other features people will like. Most importantly, it enables you to fly a flight once and then use this flight as a basis for future autonomous flights.

For example you have someone who wants to inspect farm fences from the air or maybe a cooling tower etc. What you do is fly the flight they want in record mode, you have now set your waypoints that can then be replayed. Of course you can also clean up the recorded waypoints in the ground software as well but the R&R technique is just another way to set waypoints for a flight than using the GCS.

R&R is specifically aimed at physical infrastructure inspection jobs where an expert flies exactly what the client wants to see as this is much faster and easier than guessing on a map. Once the flight has been performed, the client then has a flight he can replay without the expert being present.

The main different OpenPilot brings is we started from day 0 as a UAV, it is not a Quad platform, that is just a type of airframe we fly, the architecture is designed to be a high-end UAV.

Finally I want to stress something very important: The fact we sell the hardware so cheap is a political choice, the fact the software is free and Open is a political choice. Our cost is leading people to compare OpenPilot (and other Open Source projects) with projects that cost a similar amount but that are commercial. This is a huge mistake, the commercial projects charge exactly what they cost to develop plus profit, you spend $400 on a commercial project, you are getting what you paid for minus the profit as the developers want to be paid for their time.

OpenPilot and other Open Source projects are different, the developers, documentation writers, testers etc give their time for free to build something awesome for everyone. Not for a profit but to create something fantastic, something that one or two people could never hope to achieve but as a community and working together we will.

This is why we have the commercial guys running scared and if I was in their shoes I would be as well. I've said it before and I'll say it again, I am in awe of the talent involved in OpenPilot and these guys are doing it just for the love of it and for the community.

01:21 AM

RCvertt

April 05, 2010

Quote:
Originally Posted by G-unit
Man Ken,
You sure are getting good,
I'm J

I still say that vid is too good not to be from a simulator.

I can't fly like that but I can hover in one place better than anyone else so if the beta hardware is free count me in.
I'd pay for it but I've already been saving my pennies for a UAVX.

Quote:
Originally Posted by jamied
...i feel that the cost for the beta hardware should be on the user...

I agree.

Except that I can't justify the cost right now so I disagree

Last edited by RCvertt; Apr 05, 2010 at 07:34 AM.

06:34 AM

dankers

April 06, 2010

Quote:
Originally Posted by RCvertt
so if the beta hardware is free count me in.

Sorry but this is the kind of stuff that makes me want to just pack this in.

I don't want to go off on a rant here but some people are simply not at all getting the ideas and philosophy behind OpenPilot. This is meant to be a community about sharing, friendship and having fun while making something cool as a team.

The *only* reason we sell the hardware is to keep the costs down and the quality up. Here's are some facts:

1. So far I have given over $12,000 to this project, that US$. It's a gift to the community to start things moving. I don't think people here have any idea how much this all costs, just today I got $300 worth of boards we can't use and need to be re-bought, this happens a lot.

2. I have spend around 14 hours a day, 7 days a week for the last 4 months on this. I'm not the only one there are many people that puts hours of effort in to making OpenPilot awesome for you guys and all for no pay and expect nothing back.

3. There are around 20 boards out there now, every single one I gave away and every single one I hand built which takes ages. With every single one of these boards except one I gave away a floss-jtag as well. That's well over $2000 worth of hardware, all from my pocket.

4. I have always shared the knowledge we have gained from this project in a hope it helps other people. For example to get the best GPS for the project I spent over $1500 last week to do all the testing plus all the time myself and Angus put in. We are sure we have the best GPS solution out there but what do I get for posting our findings to help others? Just tired old over priced commercial projects that did not do any testing claiming we are wrong because they use a poor quality unit. Instead of a thank you for helping them improve their products from our work and expensive, we get this shoot this messenger junk. Just so people know our GPS kicks the ass of everything else out there simply because we put a lot of time and money in to make it. It is a custom designed 10Hz unit with an active patch antenna and running customised firmware that is perfect for all UAVs, fixed wing and Quads.

5. We get people ripping off our ideas before we even get OpenPilot released and we get commercial projects just lifting stuff straight from our website to claim as their own.

I am tired of people that just want to take, I am tired of the dirty sneaks that send everyone that post encouragement or defends the project here snide PMs about how evil we are, this is a totally altruistic project trying very hard to always do the right thing. Don't like it? Think we are all communists or will hurt your profits? Fine, don't read this thread but actively trying to damage this project behind the scenes with lies is just a whole new level of evil.

We have had people that don;t understand trying to buy the project for a lot of money, we have people threaten us with legal action if we export it to Europe with the autopilot features enabled (screw them, they can try, we are changing nothing and don't respond well to threats).

Last week I had a guy tying up my time asking me loads of questions on our forums via PM, not about OpenPilot but about MikroKopter as he is starting a commercial business. I spent two hours answering the basics that can be easily be found elsewhere but it got too much and I told him I did not have the time and pointed him to the MK dealer in Australia. His reply: "You might find if you give a little it comes back to you". WHAT!?! Unbelievable and there are many other self serving, self entitled people I have come across just like him, he's the kind of guy that will go around feeling hard done by and slagging off the project because of this for sure.

Let me be clear: this project rocks, the developers in it are outstanding, the hardware is outstanding and the aim is purely to do something great for everyone with no other agenda. Everyone involved is a perfectionist and as we are not commercial there is no need for us to rush things, it must be perfect.

The hardware is sold as close to cost as we can, we are not a business and from the start we have made this clear, there are no special deals or price breaks because everyone gets the best possible deal.

I didn't mind funding hardware for the first developers out of my own pocket, not only do I appreciate every second they help with OpenPilot but the hardware could have been faulty as well (it wasn't). Plus every single one of them that are actively working on the project offered to pay for it as well.

As I said on the Podcast [ http://diydrones.s3.amazonaws.com/mp3/diydrones%20-%20episode%2016.mp3 ] this is somewhat of an experiment to see how people react to being given something awesome for a low price, would they get involved and actually help the community or would they just take from it? There is no way I am giving away hardware out of my own pocket for people that are going to be testers, not only can I not afford it, this is just against the idea of helping the community. The hardware sold will never be beta, all hardware ever sold will be final and well tried and tested, the software will be beta but most of the money has gone in to making sure the hardware is totally correct.

OK, rant over but I am not sure a lot of people realise this and the stuff we have to put up with, this is not aimed at RCvertt at all, he mentioned beta hardware and this is something I would never release, I agree Beta hardware should be free even if I have to pay for it but it needs to go to people who are embedded developers. I just think it needs to be said so people are aware of the effort the whole team puts in and I wanted to get it off my chest as it have been annoying me for ages.

There is a lot happening behind the scenes and I can guarantee the project will be even cheaper thanks to some very nice research by one of the members of the community

Finally, OpenPilot is not me, I am speaking for myself here and just me, from my point of view! I do NOT speak for OpenPilot as OpenPilot is a community of around 30 people now. The project is the community and belongs to the community, I'm just basically the janitor of the project and one of the least skilled members by a long way. It is hard to keep up when we have team members who wrote software for the autopilot used on the Airbus A320.

Anyway, even if I give up (which I will never do), the project would be just fine, it already has a life of its own and I am very proud that this has happened.

Last edited by dankers; Apr 06, 2010 at 03:57 AM. Reason: Can't spell.

03:46 AM