Thread Tools
This thread is privately moderated by Maksimuzz, who may elect to delete unwanted replies.
Nov 17, 2018, 02:09 PM
Registered User
Thread OP
Discussion

Autopilot on AT91SAM7: Flight tests


7.4 15.07.2017

I had return to work on the autopilot by 2016 only. I decided to finish the job and install the GPS. So, when GPS processing was written, it was necessary to check everything. Now I want to introduce my Dusty. The layout is made on the basis of the Dusty toy. The ailerons and rudder were already moving,

Name: DYSTY.jpg
Views: 45
Size: 898.4 KB
Description: Layout of airplane

I had to cut only the elevator. In the layout 3 micro-servos and one ESC are installed. What is it for? To debug the work of the controls, I did control races on the car and walks with the layout, to test the flight algorithms using GPS. It is very difficult to understand if everything works correctly, and the layout immediately shows how exactly the steering gears work out and how the planes of the aircraft move. Also recorded a short video demonstrating the work of Dusty. After debugging on the layout, the autopilot was installed on the aircraft. In the test dates directory, you can also find a log of the journey on the Dvinskaya-Scottish-Nevelskaya route, and the route from the Michurinskoe-Dvinskaya. The latter contains 25 waypoints, and this route was passed correctly, which showed quite an acceptable system for deciding whether to reach the next waypoint.

7.5 09.08.2017

It was the first flight of the "Dragon" with the GPS module, 5 years after the first flight. The battery was the same. A fragment of the flight can be viewed on the video. On this day there was a strong south wind about 20-30 km / h. The approximate flight path is shown in Fig. 23, which was calculated according to the black box. The aircraft took off on the south (down in the figure), then it flew a course of about 320 degrees, but then (yellow line) the plane began a long reversal. When entering it, the GPS speed reached 90 km / h, air speed was about 60 km / h. Part of the arc to the intersection with the line of the expected landing, he flew 280 M at an average speed of 75.6 km / h in 14 s, this can be calculated from the flight data. He flew the rest of the 684M arc in 36 seconds at an average speed of 68 km / h. Such an arc came out most likely due to strong wind. In this experiment, the motor was controlled from the remote, because I kept myself able to turn off the engine if something goes wrong. As soon as I decided that the plane was lost, I turned off the console, the plane boarded the forest. It can be assumed that he could fly to the landing site, it is clear that he performed a U-turn, but with too large a radius.

Name: Find09082017.png
Views: 4
Size: 1.22 MB
Description: Track of flight with GPS whit throttle control from console.

In the video, a fragment of the flight shows that the plane swings its wings. Later it became clear that the reason for the weak attachment of the wing to the fuselage. Still, according to the graphs,

Name: Veschevo_09082017.png
Views: 3
Size: 96.4 KB
Description: Black box data

one can estimate the wind speed and its direction in this case, about 180 degrees. During takeoff, the GPS speed is significantly lower than the airspeed. It is also seen that the airspeed is calculated quite close to reality. You can also see that the speed measured by GPS is about three seconds behind the airspeed. This is due to the fact that to calculate the speed of the GPS module takes 2 seconds, and another second to transfer data to the controller. Waving wings, too, can be seen on the chart Roll. The graph CurrentHDG shows that from 71 seconds of flight a linear course change begins at a rate of 5 degrees per second, this corresponds to the arc of a circle. After the construction of the graph, it became clear where approximately it was necessary to search for the plane, since, as it turned out, the trajectory of the flight was really hard to see from the ground. After a long joint search, the plane was found by chance by my sister (after 6 hours approximattely). In addition, a change was made to the electrical circuit, which is not shown in the documentation: a self-resetting fuse (2.5A MF-R250) was inserted into the autopilot supply circuit, because when the voltage drops below 3.5V, the regulator tries to maintain the power, that leads to overcurrent. This can lead to the burning of the wiring when the polymer battery is deeply discharged. Fully all the information can be found in the test folder.
Last edited by Maksimuzz; Nov 29, 2018 at 12:48 AM. Reason: Vodeo was translated to english. Sorry for my english.
Sign up now
to remove ads between posts


Quick Reply
Message:
Thread Tools