SpaceX moved its first- and second-generation Starlink satellites out of harm’s way 207,152 times over six months. The Space Review reports the figure, citing a SpaceX filing with the US Federal Communications Commission (FCC). The company orders a maneuver once the calculated probability of a collision reaches 3 in 10 million.
That threshold is more than 300 times stricter than the 1-in-10,000 benchmark the same filing describes as the industry norm. Probability isn’t the only trigger: SpaceX also applies a minimum safe separation distance between objects. Maneuvers in the hundreds of thousands are nothing new for the constellation, which has previously logged 355,000 collision-avoidance moves in a single year.
Still, the probability of collision, known as Pc, is only an estimate, not a direct order to change orbit. It depends on predicted trajectories, how precisely each object’s position is known, tracking quality, object size and the geometry of the encounter. NASA handles these calls in stages. It first flags potentially dangerous conjunctions, then assesses the risk, and only then decides whether to adjust a trajectory.
The trouble starts, the author argues, when that whole chain is handed to software. The program receives a conjunction alert, calculates Pc, checks it against a threshold and draws up a maneuver plan on its own. At that point, a probability figure effectively becomes permission to change orbit.
Follow Mediasat on Google News: the latest in TV, telecom and tech
Every maneuver comes at a cost. It burns fuel, can interrupt the payload, reshuffles future conjunctions and may require fresh coordination with other operators. So the same Pc value can call for different decisions depending on how fresh the data is, whether the other object can maneuver, whether a crewed spacecraft is nearby. How much time there is to coordinate and how much fuel is left.
At the European Space Agency (ESA), a collision probability above 1 in 10,000 typically puts several teams to work on a maneuver plan. Crossing the threshold, however, doesn’t change the orbit automatically. For autonomous systems, the agency proposes separating three layers: assessing the situation and its uncertainty, making a decision that weighs the consequences, and holding the authority to execute. Small, reversible corrections could be pre-approved within set limits. Major orbit changes or moves near protected zones, by contrast, would face stricter conditions and coordination.
The issue only grows as constellations expand, and SpaceX is adding more: it plans to launch Starlink V3 satellites on Starship’s 14th test flight. In August 2026, the US Traffic Coordination System for Space (TraCSS) had 70 major users. The Commerce Department is developing the system under Space Policy Directive-3. It covered more than 11,345 satellites, along with government programs from 10 countries.
According to The Space Review author, such systems should keep observation separate from calculation, and recommendations separate from authorization to maneuver. Alerts, the author writes, should come with details on how old the observations are, which model was used, the covariance (the uncertainty in an object’s position and velocity) and any assumptions made. Otherwise, with thousands of autonomous spacecraft in orbit, one satellite’s maneuver could alter predictions for others and trigger a cascade of new close approaches.
