I have been successfully executing autolanding for a while, but I still don’t fully understand how the logic works. The new addition of TECS_LAND_ARSPD raises some new questions and brings up older ones.
TECS_LAND_ARSPD - [quote]When performing an autonomous landing, this value is used as the goal airspeed during approach…[/quote]
When does this parameter take effect? The problem I experience with autolanding is that the plane will aggressively roll to its 60 limit on the base to final leg. While this aggressive behavior is needed to stay on the mapping flight lines during the mission, it will cause an accelerated stall close to the ground when I am slowed up to approach speed.
Ideally I would have a waypoint command called DO_SET_APPROACH which would activate a set of approach parameters to limit bank angle, adjust waypoint radius, set flaps, and set the TECS_LAND_ARSPD.
For LAND_FLARE_SEC the wiki says: [quote] Time before landing point at which to lock heading and flare to the LAND_PITCH_CD pitch[/quote]. However it also says that LAND_PITCH_CD is [quote] Used in autoland for planes without airspeed sensors [/quote]
So does this mean that LAND_FLARE_SEC is only for non AS equipped planes? If it applies to planes with AS, then see question 3…
What is a “flare” for an airspeed equipped plane? I assume it means the throttle will be cut and there must be some minimum pitch that the APM will allow to avoid a sudden dive to maintain airspeed. Or does it lock the pitch totally?
Does the waypoint radius of the LAND waypoint have any effect on the flare location? Or is the flare purely defined by LAND_FLARE_ALT and/or LAND_FARE_SEC?
Were there any other changes made to the code in 3.0 that affect autolandings?
The wiki is pretty vague on this complex subject.
Thank you.
I fly on a very short asphalt parking area that is inclined and has a curb opposite the landing direction
I have never tried auto landing because it would have to be very precise for my situation.
Fleshing out the Wiki some more on this and a few other areas would be a big plus for those trying to get the most out of the APM/Pixhawk product.
To add, is there a way to incorporate sonar into the mix for a poor mans radar altimeter?
Steven
I haven’t bothered much with autolanding in the past because there is too much height uncertainty at the end of a longer flight due to baro alt drift, which combined with a 5 degree glide on power cut means up to 70 m of variation in touchdown distance which is too much for my flying area. At the moment, until we get as solution to height uncertainty, the suitability of the autoland for a given user is really dependant on a combination of the length of landing area, glide slope angle of the model, flight time and airframe robustness (it doesn’t control sink rate at touchdown).
I and others have tried to use sonar units for planes without much success. There is too much Doppler shift (positive at the front of the beam, negative at the back) with planes due to their relative speed over ground, which smears out the frequency of the returned signal, and reduces the amount detected by the receiver, which means that over grass, they become useless.
However this is about to change as there is work going on at the moment to incorporate range finder (and later on optical flow data) into the EKF which will provide a reliable estimate of height above ground that the plane can use during landing. There are also some miniature range finder solutions, one in production and one about to start production that will provide a solution,
In answer to the questions about the landing behaviour, my understanding - Tridge feel free to chip in if I’m off the mark - is that:
The Plane enters landing ‘state’ once it is in-bound to a landing waypoint. This is when the landing specific TECS parameter gets applied.
When it gets to within either LAND_FLARE_ALT of the landing waypoint OR within a LAND_FLARE_SEC of the waypoint, it performs what is somewhat euphemistically called a ‘flare’
during ‘flare’, throttle is cut, pitch is controlled to LAND_PITCH_CD and the bank angle is limited to ± LEVEL_ROLL_LIMIT
The Plane enters landing ‘state’ once it is in-bound to a landing waypoint. This is when the landing specific TECS parameter gets applied.
When it gets to within either LAND_FLARE_ALT of the landing waypoint OR within a LAND_FLARE_SEC of the waypoint, it performs what is somewhat euphemistically called a ‘flare’
during ‘flare’, throttle is cut, pitch is controlled to LAND_PITCH_CD and the bank angle is limited to ± LEVEL_ROLL_LIMIT[/quote]
Paul, Thank you for your response, however it still didn’t clear up some of my questions.
You said the during flare the pitch is controlled by LAND_PITCH_CD, but the wiki says that is only for planes without airspeed sensors. Does it apply to AS equipped planes too? If not, then how is pitch controlled?
For airspeed equipped planes do both the FLARE_ALT and FLARE_SEC apply?
During ‘flare’, pitch is set to LAND_PITCH_CD regardless of whether there is an airspeed sensor used and the criteria for entering flare is the same (based on either time or height, whichever occurs first).
During landing, but prior to ‘flare’, there is a difference depending on whether an airspeed sensor is being used. If using an airspeed sensor, the pitch is allowed to vary between the normal min and max pitch limits. If not using an airspeed sensor, the pitch is constrained between 0 degrees and the normal max pitch limit, presumably to prevent excessive speed building up if the glide-slope is too steep.
Michael Day’s DIYDrones blog post on his new Rally Landings code contains the best description of all the autolanding parameters, as well as a view of what is coming in the future.
Until the Wiki is updated, this post contains the most accurate information on the subject: diydrones.com/profiles/blogs/lan … e=activity
