I just downloaded the CSV to get all the info; until now I was just looking at the summary generated by healthydrones. From the summary I thought it took off with 3.6V which could never correspond to 50+%. But from the csv its clear he took off with ~3.85v per cell (on the ground, no real load), which could well be in line with 56% capacity.
So there are several problems here; if we assume the voltage readings to be correct, then the battery drained way too quickly for some reason. The most obvious explanation is a faulty battery. Bad batteries happen, thats a fact of life, but that would still be a DJI problem. The only other explanation I can think off is a short somewhere, leading to excessive power draw.
The second problem is much more crucial: the battery state estimation clearly doesnt handle a bad battery or excessive load situation. It appears to rely on a calculation that takes initial battery state before take off (no load voltage) , then deducts measured (or estimated) current draw. That only works if your battery condition is good and known. But regardless of the actual algorithm, or if the root cause was with the cells or an incorrect voltage reading, if you are reading ~3v from all cells, the software simply
has to know there is an urgent problem (and that capacity could never be >50%). There is
no excuse for not handling this (and taking off with a full battery will not always prevent it). Most RC pilots will have those $3 lipo alarms that will beep if any cell drops below 3.3V:
Vertigo, Beware that at lower voltages the internal resistance of the battery is much higher, so that on load the voltage drop is much larger, and if the drone has not flown for a while the battery is colder further increasing that internal resistance. DJI recommends that the battery is fully charged before flying to avoid explaining all the details of using partially charged LiPos. Also if one stores LiPos at too high a temperature, or at either too high or too low voltages causes the LiPos to permanently go to higher internal resistance values, further aggravating this situation, which is best avoided by never flying to low reserves of battery capacity.
That a ~$1K RC craft lacks this most basic protection is beyond me. Stop calling this an intelligent battery, its an utterly retarded one.