Today we were having a discussion about aircraft control via different flight modes. Personally would prefer to change the way input happened when controlling the UAV directly, so that the input is more intuitive in nature. I have seen allot of really well designed controllers recently however little that changes much of the traditional levers and dials, and nothing a single puck. Maybe there is something to that Star Trek control. Above is just a quick conceptual collage. The base was squished as you can see the original here, and was quickly modified to stimulate thought on how a typical RC twin 2 axis stick design could be evolved so that a more intuitive control can be developed. The center puck above is a true integrated 6 axis control, all that can have coordinated intuitive movements for the UAV. 


Here is some of the verbiage from the 3dconnexion.com:

"Superior 3D Navigation

SpacePilot Pro’s patented 6-Degrees-of-Freedom (6DoF) sensor is specifically designed to manipulate digital content or camera positions.

Simply push, pull, twist or tilt the 3Dconnexion controller cap to intuitively pan, zoom and rotate, while your other hand simultaneously uses..."

Ordered one today to take it apart, and see how I can hack together a solution, possibly integrating a board computer and a small wifi-bridge/ap so that it can be used as a integrated base station. 

It seemed obvious since the object we are controlling actually operate in a 3D environment that the controller especially when operating FPV or "First-Person-View" should be able to translate that intended movement as turn left right, up down, and with a few that could be atypical due to the way and coordinated function of the device and its ability to tilt or go sideways, possibly to engage linear thruster type dives. 

The Myo and the 6-Axis mouse will be a fun combination, it will be interesting to integrate tactile and and virtual gesture based movements. 

Dev Team



E-mail me when people leave their comments –

You need to be a member of diydrones to add comments!

Join diydrones


  • Sam, very interested, also in replicating your setups here...

  • I would love to help you guys test but I am in Australia.  Have you set it up for Southern Hemisphere operation.  Everything spins backwards down here.

  • Alex,

    We are shortly getting a small patch of products put together for field trials. If you are interested, we could probably send you a Wing to test?


  • Thanks for the reply.  I agree that the setup of the controller is the key to getting a good feel to the system.  I remember trying to fly on stock setups with no expo or limits, not so much fun!!!

    I do think that no matter the style of controller, if it is able to provide intuitive control and possibly feedback then it will work.  I must congratulate you guys on trying to come up with a different approach and I hope that it works out.  I hope I get a chance to try one in the future.

  • Hi Alex,

    expo, dual rate and °/Second or meter / second are exactly the same as on a standard transmitter.

    The feeling is a bit different, we have a tweak on the system, but in fact is very intuitive as if you release the sensor, well it's like all controls on a standard transmitter are back to neutral.

    For example, it is much nicer to fly a Phantom with the joystick that with the original transmitter. reason being, you setup attitude gain in the DJI assistant software which is the same as setting up dual rates on a transmitter, which you can not do of course on the stock DJI transmitter. Then, you can modify the deadband and the exponential.

    I have seen so many people buying " expensive" rigs such as S1000, S900, and when they come at the workshop to ask for help because they can not " feel" the machine in flight, we discover that except the Attitude gain there is  nothing setup on the machine or the Zenmuse gimbal on the transmitters. 5 minutes of simple dialing ( Expo / dual rate , for each flight mode, throttle curve for changing from attitude to manual without having the machine jumping or dropping on throttle, and End points / dual rate / expo on gimbal Tx and the customers feel like they have a brand new system in their hands, much more comfortable to fly..

    This is exactly the same with the 3d Sensor : you have to adapt each axis sensibility to make it natural. It has always been a kind of " compromise" between mechanics and electronics for any Rc model or UAV ( a UAV is a Rc model called Drone or UAV, must be better for marketing lol...) . If the mechanics feeling on the input does not reflect or transmit the behavior you would like the machine to have, then you can have the best system in the world, it will simply not be matching our expectation.

    So what we do is that we setup a "base" setup on which most people feel confortable, but then we modify it for each pilot needs and feelings.

    If you need to twist the sensor like crazy to get the machine flying the way you want of follow the trajectory you wish, it will be the same that if you have to move your 2 Tx sticks like crazy to manage the 4 channels (Ail, Elev, Throttle, yaw) : your fingers coordination and extensions in the movements will become.. erratic and not accurate. Have you ever tried to make a vertical climbing followed by a spiral ( elevator / yaw / roll input) at perfect constant rate ? That is not that easy and results will change depending on your flying skills but also on the TX settings.

    Finding the right match on the electronics settings ( sensitivity, dual rate, expo...) is exactly like seeing up your seat on your car to feel confortable with the pedals and the steering wheel. Once you get this solved, everything else is much easier to handle and more " natural".

    What we found out is hat most " beta testers" do not use the yaw so much except to change the orientation for the camera once they find the right spot to look at, or will use the yaw but alone for some time and then go back to cyclic and throttle ( altitude should we say..) control. But what was very " educational" was to see how people simply natural " decompose" the action they want to perform. As the 3D sensor is just intuitive, they move from an action to another ( for example climb, stop, turn left, move ( roll) right, move forward, descent a little, turn right, backward..) in a decomposed pattern and do it very naturally. Then , after some practice, they start to mix the different inputs by themselve without making major mistakes very quickly. If you give them a transmitter, they mix inputs not by willing but by the lack of coordination and assignment of each control on 2 separate sticks. They wanted to move forward but climbed instead. They wanted to turn left but rolled left instead.. and so on. This is confusing for them and they finally spent more time focusing on how to give the right input rather than  checking how to control the machine. This is what we could observe, but maybe i'm wrong, i'm not a cognitive or psycholmotrician specialist, but i asked a psychomotrician doctor and he liked the intuitive feeling of the 3D sensor, because simply no particular coordination learning is necessary at all to handle it. It's as easy as rubbing an object and moving it, like the cube demo software for the ones who know it :)

    To answer your question fully, the throttle / altitude is very easy to handle, since DJI and Mikrokopter ( for 3DR i need to test) now do not take off in Barometer (MK) Attitude or ATTI/GPS mode (DJI) as long as you do not move the throttle over 50% ( neutral). So in a few words, you launch the motors with press of a button ( simple 4 channels instant sequencing), and as soon as you will pull the sensor up, machine will take off, at a height of 50 cm or 1 meter already barometer is working fine enough to control perfectly the altitude. It would be much harder if we did not use the barometer / altitude control, this is a fact. But flight controllers give us that " ease" and using it is just so perfect i have to admit. So once you are hovering, except small corrections or altitude changes needs, there is not a lot you need to do, and playing with cyclic or rudder does not affect throttle with the right settings and after one or 2 flights to get used with the system.

    In short, when you leave the controller at it's neutral well altitude changes are very small with a well setup and well balanced machine, so you can sue line of sight or onboard FPV or simply the telemetry to check that altitude. Same as on a standard multirotor i would say, when using altitude control of the flight controller.

    Once important thing we found out is that for " nose in" flight, unexperienced pilots were more confortable with the 3D sensor than with regular transmitter ( inverted inputs). I would guess that it's because there is only one input so inverting is easier to handle ( you rotate something for the yaw, you do not give a " stick side input", and it seems they "ink" roll and yaw kind of easier with he 3D sensor for nose in.

    In fact, if you look at the " mass" customers market all using something like a DJI Phantom do NOT know how to fly without the GPS and IOC... So many people fly multirotors only because of these features. This is also the main reason why Drones / Multirotors now come more and more to government / civilian security / army usage, as they are much easier to fly than Rc helicopters. Self leveling was the first real " revolution" in multirotors. Remember the Intellicopter, back in 2004, and check who they have become since. It was my first contact with a multirotor in paris, and i performed the demos for the french distributor. It was soon easy to fly compared to our Rc helicopters or the large coaxial UAV's we were flight testing at that time.. GPs and Intelligent orientation control then arrived with DJI ( but the first true autopilot allowing to fly a RC helicopter in such an easy way was the Carvec System from John Cunningham, the most safe and reliable system i have ever seen).

    Add lipo batteries ( flight time), brushless motors ( power and constant efficiency), field oriented control ESC's ( we work on such things but that's another story. and someone on this forum works already on a very interesting concept....) , simple mechanics ( multirotors) and you get a simple, efficient, and cheap platform that can be flown by virtually anybody.

    The 2 most solid systems i have ever flown so far ( but i have not had the time yet to fly the 3DR, i will take that time over the week end) are the Carvec system and the " Box" from Steve Webb (check my vimeo channel with the small F550 and the black box on it) which has become then the Movi controller and that will be the Synapse i guess later on... 

    Now my guess is that beyond the input controller, the most important thing is the GPS and positioning system safety, actually i think that industry is pretty far away from something totally reliable  , but many progresses have been done and coupling accelerometers datas to GPS datas, if done correctly, will allow in the future to get an extra positioning or return to home greater safety. Spectral analysis ( stars) for night flight, sonars, lasers, cognitive systems ( TESC in France by captain Gigean from Army, the greatest anti collision system in the world, truly..) will lead to more and more safety and input controls will need to be as simple as possible. 

    This is on a ethical point of view a good thing for a bad thing, as if anybody can fly anything with almost no training, risks are exponential in case of electronics failure or simply wrong intentions. This is why FAA is trying to ban UAV usages ( a lot of fear from bad intentions with such easy " tools") and it would not be surprising that in a close future, Google and major actors of real aircraft transponders work jointly in order to have kind of " serial number" and traceability of each machine in the air ( will require users acceptance to transparence and registering, will not be easy.....) for better skies management and improved safety.

    At our level, we focus on the input control, it's a small step compared to all the things that are evolving very fast now, but i think it's in the right "' loop" direction ( well, i hope lol).

    I do not say we invented the wheel, some other people might have had these ideas since a while ( well the initial post here showed we were not the only ones working on an alternative control system) , we just try to keep the last year evolutions in term of flight controllers moving on with the way to control them. But there is so much to do :)))

  • Fabian, It sound like it is going well and I agree with your comments about new vs old style users and controllers.  I would love to try the 3D controllers as I think it would be great to fly missions and general no acrobatics flights with it.  I bet that after a while, once the setup is really dialled in that ACRO wouldn't be much of a problem either.  You are right, it is all about getting the control parameters correct. Do you find that things like Expo urges are different than for sticks?  How sensitive is the throttle/altitude control? Do you get much feedback from the controller to determine throttle level or do you find that it is more about visual feedback from the platform (ie FPV Camera or line of sight)?

  • We are glad this thread is moving the right direction and of course we are happy to read these nice comments.

    I say " We" because my ideas would have gone nowhere without David and Tiago and also 3D Connexion, who supported us with the hardware and a few goodies ( we broke a few keyboards and sensors lol) since the very beginning with Raik, Uwe and Antonio now at 3D connexion, it's a team work since the beginning and my only hope is that all the hard work these past years can lead to something concrete soon for all of the contributors in this project :)

    I will try to post a video of the system with the 3DR flight controller as soon as possible :)

  • Tearig, I agree. The attitudes on here are really positive and it's really good to see alternative approaches towards the same goal.
  • Sometimes the universe suprises me this time is no differnt, the distance covered in just the short time since I made this post has realized so much benift and I am so grateful to all that have contributed so far! 

  • Hi,

    there is one variable that we did ignore for a while and it was an important one, when we started the project : People used to Rc transmitters ( standard ones) like us ( i fly Helis and planes and of course multirotors) have the "coordination" set in their head already, so it is in fact harder for them to use the 3D sensor at the first time. This had to be put aside when we made the tests with "normal" people with no flight experience.

    If you want to fly artistic or 3D, then yes, maybe a Rc transmitter could be better of course. But if you want to fly a " Tool" or simply follow an object or a path, then, the 3D joystick is way better, because it's simply like moving an object in space.

    Throttle or pitch are used normal to stabilize altitude. But if Altitude is stabilized with a barometer, or a Gps, then you do not use it that much for mapping construction sites, fields or survey. 

    Standard mission is, in most cases : control the machine during hover and get it into position, then switch to auto pilot or use GPS or self leveling ( compass or accelerometers) assistance, use barometer assistance, and move the UAV along a simple path to observe, take pictures or others. Add Carefree / Intelligent orientation control, and honestly, this is just as simple as that to fly a mulitrotor or any Vtol equipped with a decent flight controller.

    Most people flying for survey / observation or mapping do not use a lot the throttle except when taking off and landing.. But Throttle / altitude control remains very easy even with the 3D sensor.

    Also, we could notice that "standard" users ( who never flew a rc model before) had all time a better coordination in the 3 channels ( rudder, aileron and elevator) when using the 3D sensor than a regular transmitter, because they didn't have to triple check each axis on each stick's coordination during the time they wanted to perform the maneuver. They did it, simply. Everything else is a matter of deadband, sensibility, exponential and flight controller setup ( Tiago Inside lol) and experimentation with as many people as possible.

    All of the " testers" have been invited to test the system this way : on first tests, we had trainer cord between the system and a regular transmitter. We hovered, and then placed the multirotor at 20 meters altitude around, in GPS with carefree  and gave them control on the 3D sensor. We just told them to keep it smooth and experiment, and release if anything went wrong. After one minute or 2, we switched to " attitude" ( self levelling / altitude control, but no GPS and no carefree) and helped them a little, gave them advises ( keep nose pointed forward, try to keep position...), and after a few minutes most people were already able to fly pretty well already, some of them were even able to land with no GPS control. After 3 or 4 flights they could bring the machine almost everywhere they needed to. With our 2 Super Grannies ( 70 and 74 years old) that we met at the field at CEEMA, well it was a real blast, they did not want to stop flying and even landed on their own at the very first flight. That was the coolest thing in all our experimentation, to see them so happy and enjoying flying a multirotors. Kids loved it also a lot :)

    Now we do not use trainer transmitter anymore and directly give the joystick to the "tester" , and we had no troubles so far. 

    We do it even indoor at the workshop when some people come to test it, and nobody crashed anything so far ( lucky me..) :)))

    We learn them how to fly without GPS assistance and after some time they can fly even manual for some ( full manual, no self leveling).

    If you want the to do the same thing with a standard transmitter, they need almost one week of training to do this.

    The very first goal was ( and is reached now) to allow anybody like a firefighter, a policeman, a soldier, to control a UAV or Rover for survey or observation, to fly with no extensive training with a Rc transmitter, and always keep the same " logics" with he control system. As it was pretty successful we did let interested persons wanting just to fly a multirotor testing it and well they like it over standard Rc transmitters and unlike us, they are absolutely not bothered by the multiples axis being handled in just one input device.

    Then we have modified the sensitivity to allow partialy disabled persons, especially children, to fly a multirotor, just for fun. 

    And then, well, we just moved forward :) There are 3 different versions of the system up and running actually, one super portable version, the standard one and a mission oriented one.

    It is true that it is better to have a heavy "base" to be able to have perfect control at the beginning, but well there are several ways to make things work ok..

This reply was deleted.