I was playing around with autolanding. With more luck than wisdom I found pretty neat landing parameters after 5 consecutive landings. A little overshoot: 7m. The plane should have landed at my feet)
What I did was increase flaps in steps of 10% and decrease minimum airspeed. Moved the landing start point around a bit (not very systematic). I also lowered the flare point from 5 to 3m.
Setup: Ranger 757, ArduPlane 3.5.2, The sonar is an analog Maxbotix. Airspeed is I2C pixhawk. Wind was 3bft. Shifting direction a little.
Now it's nice to be lucky, but it's better to be prepared. :) So, what would you suggest as systematic approach to tune the parameters? I want to tune for:
1. The softest possible landing
2. ... on the shortest possible runway?
How would wind impact the parameters? Does more wind require a steeper descent rate? Or will ardupilot take care of that with more throttle?
Here is my plane with pixhawk and LIDAR Lite 2. I would suggest getting rid of the Sonar and move to a LIDAR. It is way more accurate. Plus you can get the LIDAR Lite to engage at around 100 feet which would produce a stabilized approach. Mine kicks in when it turns final which is around 100 feet AGL. Its cool to see after a long flight the LIDAR correct the altitude due to pressure changes.
Flaps are used to make a steeper approach without gaining any more airspeed. Keep in mind the windier it gets the less flaps you want due to you will lose authority over your control surfaces. Typically a 20-30 degree flap setting is ideal. For mine I didn't use flaps btw. Its funny the UAV I flew for work actually put in flaps on Base and final but SLOWLY took them out on short final. It was mainly because with full flaps in on that plane you couldn't get the nose up high enough due to lack of authority of the elevator due to the flaps.
The softest possible landing is when the plane basically almost stalls once the wheels touch the ground. Keep that in mind.
Thanks! Great tips. And nice video. Nice airstrip you have there too.
It seems I have plenty of elevator authority on the ranger. With the wings above the plane theoretically the flaps should pitch the plane up. In my logs I can see the autopilot not steering up very much during the flare (RC OUT 2). You can see flaps on RC5 an throttle on RC3. Apparently the whole descent worked fine without throttle. I guess the autopilot will increase throttle with more wind, to stick to the line of descent?
Question: What does the lidar data look like? Is it cleaner than the sonar? Or is it just better range? Can you post a log or a screenshot? I'm afraid to engage the sonar early, lest it catches a tree line or a bush before landing. I've also seen people send a barometer offset from the GCS just before landing.