/sci/ - Science & Math

Hullo's summary of the Hakuto R autopsy
https://www.youtube.com/watch?v=2JlUnOAiMm4
tl;dw: edge case in software made the lander ignore the radar altimeter

>according to the team, what happened was during the approach towards the landing site the vehicle passed over a crater with steep sides, and all the time the radar altimeter is pinging this thing and it sees the altitude suddenly change by three kilometers as it passes over a steep cliff, and the software says "wait a second, the rate of change of the altitude shouldn't be this much; I think that this sensor is broken" and so like any well-programmed piece of software it began to ignore the spurious data - except that the data was actually correct.
>the spacecraft thought the surface of the Moon was about five kilometers higher than it actually was, and so the control software did a perfect job of setting this thing down at its altitude where it thought the lunar surface was and then it just sat there balanced on its rocket motor slowly descending waiting for the moon surface to come up and greet it, and before long it ran out of propellant - at that point it actually ran out of propellant asymmetrically and it began to spin, by the looks of things; but yeah it basically fell those five kilometers to the surface