A few days ago I wondered if anyone had considered adding some sort of “point of no return” fail-safe.
While a certain amount of inaccuracy may have to be accepted, it ought to be possible for the autopilot to determine when a threshold is reached where there isn’t enough power available to complete a RTL.
In fact, this might make an useful fence definition.
If nothing less, a warning might be useful.
Fail-safes such as low voltage and critical voltage give some chance of a controlled landing at a critical point. But as Copter doesn’t have rally-points, there’s no option to bring a copter back to a spot of land free of water of vegetation hazards.
Anyway - I was just curious if this sort of thing had ever come up for discussion.
b) As you’ve mentioned, an automated fail-safe system can be be a challenge if drone happens to be over forest, buildings or water.
(I had a rather expensive drone once destroyed due to “landing” being triggered by a fail-safe function over a forest)
This would be pretty interesting, I’d guess this is something that could be trialed in a Lua script fairly easily.
I think most people would just want a warning but depending on the use case I could see the value in the auto RTL too.
The main challenge would be calculating the efficiency of the drone during forward flight based on the beginning of the flight.
Doable I think, would be a good project for a student in our autonomous systems class. But it would need to wait till next year at least as they already have their projects for this year.
It would definitely be something nice to have and very useful. It might require these for variations of automatic (dynamic) or manually configured versions:
Smart battery so the FC knows how much capacity was used for the journey so far and how much may be needed for the return leg considering the remaining capacity.
Guarantee you are starting with a known-good fully charged battery and set the mahr related settings appropriate for each flight and conditions - ie: 50% used as the RTL failsafe.
Wind estimation and other factors would play a role too
Some would also point out that it’s up to the pilot to know their aircraft and investigate the area and flight path, consider setting rally points, and be aware of weather conditions (wind) - and plan accordingly.
DJI’s estimator works so well because they have control over the power system and all the data specific to each model of aircraft. That’s not practical in a general sense with Ardupilot and the myriad of variables and uncollected power system data.
And The DJI system does work very well. I tested it’s limits many times with the one I had.
I think the standard 20% and 10% rule for low and critical works for almost every LOS operation (can’t think of any reason why it wouldn’t) - so I feel like this requirement would be tied with those do BVLOS flights. In that case, I wouldn’t fly without having rally points setup anyways.