34 ms·

> SpaceX's satellites will be able to sense and autonomously maneuver away from potential collisions, Musk has said. (This didn't happen when a Starlink satelli

> SpaceX's satellites will be able to sense and autonomously maneuver away from potential collisions, Musk has said. (This didn't happen when a Starlink satellite neared Europe's Aeolus Earth-observation satellite last September; Aeolus performed the evasive action. A paging-system bug prevented Starlink operators from learning that the collision probability had increased enough to warrant a maneuver, SpaceX representatives later said.) So, the part about being able to autonomously perform evasive maneuvers is just a lie? If it isn't, why would a paging failure prevent an "autonomous" maneuver from occurring?

dragonwriter 7y ago
Conceivably, they could be able to conduct evasive maneuvers autonomously but be configured to seek control confirmation before executing maneuvers. Or, maybe "will be" is the key phrase and its like Tesla full autonomy -- a feature planned to be implemented in software with the current hardware, but not yet done.
mhandley 7y ago
Things are likely more complicated while Starlink satellites are maneuvering between orbits (as was the case with Aeolus) than when they're established in their orbits. In addition, as with aircraft collision avoidance, it's really a question of who changes course and how. If neither move, it may be bad. But if both move in the same way, it could be bad too. So likely the database Starlink uses to determine possibe conflicts also categorize those conflicts as passive (the conflicting "debris" is not under control) and active (the conflict is capable of moving). For passive conflicts, you MUST change orbit, and that can be automated unilaterally. For active conflicts, you MUST NOT change orbit until you've agreed who will move, and that requires human involvement at this moment. I think SpaceX learned a lesson from this, and they will have more robust procedures in place going forward. Eventually active conflicts can be automated, as they are with aircraft TCAS, but we're not there yet. From discussions with ESA folks, I believe there's ongoing work on with regards to such automation.

dragonwriter 7y ago

Conceivably, they could be able to conduct evasive maneuvers autonomously but be configured to seek control confirmation before executing maneuvers. Or, maybe "will be" is the key phrase and its like Tesla full autonomy -- a feature planned to be implemented in software with the current hardware, but not yet done.

mhandley 7y ago

Things are likely more complicated while Starlink satellites are maneuvering between orbits (as was the case with Aeolus) than when they're established in their orbits. In addition, as with aircraft collision avoidance, it's really a question of who changes course and how. If neither move, it may be bad. But if both move in the same way, it could be bad too. So likely the database Starlink uses to determine possibe conflicts also categorize those conflicts as passive (the conflicting "debris" is not under control) and active (the conflict is capable of moving). For passive conflicts, you MUST change orbit, and that can be automated unilaterally. For active conflicts, you MUST NOT change orbit until you've agreed who will move, and that requires human involvement at this moment. I think SpaceX learned a lesson from this, and they will have more robust procedures in place going forward. Eventually active conflicts can be automated, as they are with aircraft TCAS, but we're not there yet. From discussions with ESA folks, I believe there's ongoing work on with regards to such automation.