Good morning guys,
I am new in this forum, so first I want to say hello to all and congrats to your achivements so far!
For many nights I have now screened blogs and discussions, but I did not find a white paper with the goals to achive for a DIY Drone and which limitations will apply.
By reading all the comments I learned that there are great mathematicans in this community, either setting up matrisses for computation or programming.
What I am missing is the practical side of it.
Some of the stuff is used in not suitable inviroment or with improper combinations of data retrieving.
Lets see what our big brothers are doing so far:
Even the SAFU´s (smallest availiable flying units) like a Cessna 150 or the Piper similarities can be fitted out with Garmin IFR goodies and a three axis autopilot. This will give them the ability to fly like a drone by waypoint or wingleveling without checking for obstacles- the same what we do. The only difference is, that our hands are replaced by servos. Once the route is programmed in the Garmin, the aicraft in cruise flight, the pilot may leave the aircraft by parachute. Landing will be crash- but until then it behaves like the ardupilot shall do.
How is that achived:
The method of wingleveling is the oldest availiable technic. The ailerons are moved by the steering unit to keep wings horizontal only, no matter of drifting off course.This is the easiest to achive and is ready availiable for RC by many products on the market.
Very important is that the aileron keeps the wing level and not the rudder!!
A 2-Axis autopilot dampens oszillations in pitch axis by the elevator- still no rudder.
This setup will not hold altitude, just dampens oszillations.
To add altitude hold, a barometric pressure transducer is installed which sends altitude information by static pressure readings. This works perfectly, deviations are within a few feet. The GPS altitude is only used for GPS/ILS approaches with countercheck of pressure readings. More expensive systems will have an altitude preselect which will level off from climb and descend automatically at the preset Flight level.
Most of todays aircraft within 5.7 tons are fitted out with this system!
But where is the third axis, the rudder?
Well, the rudder is not relly needed, it is trimmed for speed to keep the fuselage in direction of the airflow and nothing else!
It is definately not used to change direction, this is done by a combination of ailerons and elevator.
First, the aileron will roll the aircraft to a preset angle. The standard Turn is called Rate1 and defined by 360 degrees in 2 minutes or 3°/sec. How much the aircraft has to bank is defined by the computer (function of airspeed) to achive this rate of turn. The bank limitation is 30°, what means no higher banks are allowed.-
Second choice is to turn the aircraft always with 30° bank.
Third: any turns higher than 45 ° will be done by 30° bank and lower banks for lower turns for inceased stability and comfort.
A 45° bank on AP is already an emergency.
Then the tilted elevator has to compensate the loss of lift due to the bank by producing more downward force (elevator flap moving up), this increasing the angle of attack of the wing, until the weight force and the lift force are balanced again. As the lift vectors are no longer vertical but pointing inwards of the turn, the a/c will start turning in a certain radius.
Hmm, still no rudder!
Why then the third axis, the rudder?
Especially jets with wings highly backswept (good for speed) have a tendency to start an induced yawing and rolling at high altitude (dutch rolling). Only this aircraft are equipped with the yaw control. Some of them are mandatory- that means that they are not allowed to fly higher than 20.000 ft with yaw computer off or u/s.
As written above, the rudder is used to keep the fuse in the slipstream. In turns it might happen that the nose veers outwards or inwads of the coordinated flight path. Here it is used to get the fuse back to a coordinated turn.
Due to moments produced by props, engine(s) and their airflows, the rudder is used to compensate these moments. Every speed needs an adjusted rudder position, to fly with low friction.
What will happen to change direction with rudder.
Saying boldly: Everything!
the rudder now forces the a/c to bank, turns at the same time and combined with the elevator the a/c achives a turn.
Some airliners are equipped with full controls, whereby the throttles are controlled by the autopilot as well.
Why that:
In level flight- due to flight caracteristics- the airsped is achived by power and trim of elevator. As higher the airspeed the less is the up deflection of the elevator and vice versa.
What happens if the a/c is trimmed in level flight and power is increased?
Some got it right -I am sure- but not all:
The a/c will start climbing with the same speed as in level flight, because the elevator is NOT for climbing but to preset a certain speed- per definition!
To descent- now you all got it right-
Reduce power and the aircraft will descend with the same speed as in level flight.
Every aircraft has a best climb speed, so the computer will select appropriate power for the given speed and climb rate when demanded to climb.
Going down vice versa.
This reduces the workload of the pilots incredibly, and belongs to safety features like stall proofing as well.
Only by knowing and understanding the pricipals of which surface will do what,someone will be able to successfully program an A/P. Just knowing how to program a mc or to fly an RC aircraft will be too less.
I will continue this shortly
b.r.
Wolfgang
Replies
I think you are absolutly right about the rudder use. I understand that this is a basic easystar model but it should be above the programming of a model with ailerons. I asked Chris the same question about the rudder use in a private message. We do autolandings in the plane and never is that rudder used until the aircraft touches down/10feet and you align it with the runway. The ailerons do all the work and the computer works on the ground track and flight path.
I also think this is the basic control for flying and from there you have to work up(no ailerons). Don't know a lot af people flying planes with only a rudder and no ailerons. Even ultra lights have it the 90% of the time. RC just gets away with it, you won't lose you life with crashing the rc plane.
My personal thought is that snaking to a waypoint comes from no aileron use and overshooting the desired ground track.
Have to get the basic control.
The best resource for new users at present is the wiki manual. Chris Anderson and Jason Short have been doing a stellar job in putting that together. There is still plenty to do.... It is easy to recruit people (like me) to write autopilot firmware and design hardware, etc. It is not so easy to recruit people to do things like write documentation and other tools.
Finally, if you look at the 2.5.1 firmware I think you will find that things are moving in a direction you like. Earlier revisions were written by guys who are not full scale pilots. They had many things which were "tweaked" to work and they did a great job sometimes by trial and error, but we are working towards having the firmware be suitable to a very wide variety of airframes and based on traditional aerodynamic and flight control principles. In particular I think you will like the new elevator and throttle control laws. (Please note that there are multiple control laws dependent on if the AP has an airspeed sensor or not).
BR
Doug
Doug, Thanks for details of your control laws for fast Elevon planes, I was able to fly with good Stability yesterday, for 100m and then landed with no power smoothly. I need to set max bank angle, rate of turn limits & test fly. Then set up throttle curve and re-test. I only had a short time to fly as Football and Cricket teams have invaded test range from 1- 5 pm
Your advocation of use of elevator for speed control and throttle for climb/descent is one way to do things. In fact it is one of the control law setups in use with ArduPilot. However there are other methods that are equally valid in appropriate circumstances. For example the airframe I mentioned above: I do not even use airspeed as an AP input. Why? Because the airframe has lots of power and will climb at 45 degrees all day long. By limiting the maximum pitch commanded by the AP to an appropriate value less than 45 I have no concern of stalls. Speed control is handled by use of a multi-point throttle curve based on desired climb/descent rate. In this case airspeed control is pretty loose, but airspeed control was not one of the goals/requirements for that project.
Goals and requirements are what drives choice of control laws, not what is taught by primary flight instructors or what is theoretically most "efficient". A 2 minute "standard rate" turn would be completely inappropriate in many circumstances. For example the last T3 contest had a winning time of ~25 seconds. Pretty hard to fly that mission with a standard rate turn ;)