Raw footage of 3 fully autonomous landings on a very windy day. The video description and subtitles hopefully explain a bit more about what the autopilot is doing. Successful autonomous landing means all the pieces on the ground and stopped moving, right? 
Is that squeeking normal?
@Matt_Harrington It’s the interaction of the wind with the camera mic. I don’t get it with other camera orientations. This is an engineering test video so I didn’t touch the audio track … the sound gives some clues to what’s going on (i.e. speed up/slow down, if the aircraft is getting rocked by a gust, or if I need to hunt through my telemetry data and see what the autopilot was thinking, etc.)
@clolsonus , interesting! I suppose you would not want to fix it then. I like the pattern of the visual artifacts from the spinning prop too.
@Matt_Harrington Right, the prop patterns are from the progressive scan CCD (common in most action cams, common outside the most expensive cameras actually.) I’d rather not have the patterns, but looking through the prop to gauge rpm and thus what the altitude controller is doing, is instructive in an engineering sense. (The goal of this video is more engineering and less entertainment.)
Great job with this…fantastic engineering, and it IS actually pretty entertaining too!
Thanks, it is still a work in progress (and you can see all my approaches came in high), but this was kind of a milestone day for me. 3 consecutive fully autonomous flights, including autonomous take off and autonomous landing. I am always ready with the manual override, but never touched it or the sticks for 3 complete flights in a row.
Impressive - As a former ILS and ground radar engineer I know how hard it is to get a full sized aircraft to do this.
A good demonstration that wind doesn’t scale!
Thanks for the kind words. Here’s my latest effort in a bit calmer winds (still bobbles around, but it is a big high wing trainer with lots of dihedral.)
Are you using a radio glide path and localiser set up?
How do you generate the narrow beam without a substantial antenna?
@clolsonus
Most servo systems tend to hunt either side of the aiming point - usually that’s why a PID control loop is used to minimise the over shoot.
It is very hard to do in a real life active situation because your buffeted so much by outside effects.
Your efforts are commendable.
Hi @Richard_Harris at the core this is all gps/barometric altimeter based. Back in the 50’s my dad was in the airforce and worked on a ‘portable’ ILS system that was radar based. Somehow the ground radar would detect the aircraft, and ground radar operators would then radio approach vectors to the incoming aircraft. I was thinking it might be interesting to try something similar but with a camera and image processing on the ground side. If the camera is located at (near) the touch down point, the aircraft should never move in the view if it is on the ideal glideslope and track. Anyway, that’s something for someday…
I am running a set of PID loops, so there is some hunting … you can especially see this in my altitude/throttle towards the end of the descending circle. On this most recent approach I had specified a lateral offset to push it further out to the right … but I put the number in with the wrong sign (dohh!) so it actually brought me closer in to the fence. If I would have just left it alone it probably would have hit the runway dead center.
The system should account for cross winds and properly crab. The problem in real life is when you get down around the tree line the winds can be gusting and rolling off the nearby objects, and can change dramatically on the final leg…
Just for fun, here’s a video I made in FlightGear of a precision carrier approach system I was simulating. This isn’t doing any image processing, just showing what the view would look like if the camera is placed at the target touchdown point looking up the glideslope:
@clolsonus
Reminds me - When I was in the air force the Radar and ILS system had to be calibrated every year by running a C130 down the approach path - on the end of the runway a (poor) small group stood with a theodolite and radio checking the aircraft’s approach angle.
Having a C130 pass over head at about 50 feet is beyond exciting. It did however focus the mind on getting the set up correct!.
One year at RAF Honnington we were offered a chance to ride the aircraft during the calibration runs. Unfortunately it registered a minor hydraulic fault after we took off and the pilot declared he was going to stay in the air until he could go home (to cottismore some 150 miles away) because he had a dinner apointment and didn’t want to get stuck at Honington.
We were in the air for just short of 8 hours - never went more than 20 miles from the airfield until we finished and turned for Cottismore and a long bus ride home late at night.
My wife got to ride in the cockpit of a C130 one time for an educator’s outreach event. I was jealous!