Welcome to the Robolink Community forum!
Inconsistent coded flights
How can I ensure repeatable coded flights?
I am very particular about:
-no debris on propeller shafts
-propellers in as new condition
-current on all firmware and software and library
-still air indoor space
-fresh, fully charged battery inserted every time code is flown
-drone bumper is seated properly
-motors have seen relatively little use and seem to be fine, but probably need trim adjustment because even after running the calibration routine, some drones still drift
And what sort of accuracy does a coded flight have (given that no trim is needed)? Can I expect a drone to fly the same course over and over to within say +/- 250 mm or is that unrealistic?
How many minutes of flight time are motors good for, before needing replacement? I am thinking in terms of a preventive maintenance replacement schedule. A log of flights could be kept.,
robolink_wes last edited by
Hi @javiste! We just finished our call, but wanted to list this out here for others as well.
All the things you listed are definitely good things to look out for and implement. One more thing that will cause the drone to drift is if it hasn't been calibrated for the environment. Lighting and the pattern of the floor, if very different from where it's been calibrated, can cause the drone to drift, especially if you flew the drone elsewhere.
Here's the tutorial about recalibration. Here are the steps, just to clarify.
- Insert the battery
- Hold the drone updside down
- Press and hold the side button until the drone's arms blink
- Immediately place the drone on the ground and stand clear for it to take off and run the calibration
One thing to note, make sure the surface is not too dark or not too glossy/reflective. The optical flow sensor works best on a carpeted surface with a distinct pattern. This will help the calibration process.
We would say expecting a margin of error of about a 2 ft x 2ft square is reasonable, or about 60 x 60 cm. And definitely take into account cumulative error. So because it's using dead reckoning (i.e. doesn't have sensors than can tell where it is positioned in space relative to the room), the more it veers off or changes course during a flight sequence, the larger the error will be further along the sequence.
So we tend to keep our obstacles relatively simple, like a take off pad, 3 hoops, and a landing pad. For more complex courses, we'd recommend mixing portions that are remote controlled, and portions that are autonomous.
Hopefully that's helpful!