Thread by @ainvvy

Spent the last three hours trying to understand the architecture of the lunar lander.

From the face of it, if it is just a communication problem, the lander still should have descended correctly because the lander navigation should have been autonomous.

Now, I do not how how autonomous this system is. From what I read, it seems the lander had a direct link to telemetry/tracking/control (at Bayalalu near Blore is my guess).

This is the overall guidance architecture. It is clearly IMU based. All other sensors are seen as aiding the IMU in various ways.

Most of these are for drift correction on the IMUs and get a decent localisation for descent.

Now anybody who has used an IMU would know it is quite unreliable except for a short period of time. Accelerometer is noisy and when double integrated to get position, the noise itself gets integrated and so the position can be way off.

Now in the autonomous ground vehicle I am building, we do have an accelerometer but we don't depend on it much. We depend on other (better) sensors to localise.

ISRO seems to have decided to heavily depend on IMU as the descent is only for a few minutes.

Nevertheless, even if comm failed, the lander should have landed safely. I would have expected the orbiter to have eyes on the lander, rather than just getting a pose back on telemetry. Maybe they had it, I have no idea.

If the lander did land, the rover may not have deployed because this may have depended on a manual command from earth to deploy. If not, even the rover should be roving around happily -- if this entire design was truly autonomous.

Did communication alone fail? Surely the lander talks to the orbiter even if its link to earth failed? These I could not find answers to; someone who know better can shed more light.

Now, we have had our share of failures in building our autonomous system. It is almost always some trivial programming mistake: usually some edge case not thought through. Or some wrongly swapped wires. Once we know what the error is, it is usually a facepalm moment :)

This is how all engineering progress is made, one bug at a time. My sympathies are completely with ISRO, and that first person who facepalms when realisation hits them!

Here's to many more journeys ... which is the only thing that matters.