An illustration showing the close approach that ESA’s Aeolus satellite would have made to a Starlink satellite on September 2, which led to ESA performing a maneuver to shift the orbit of Aeolus. (credit: ESA)
The controversy started, like so many these days, with a tweet.
On September 2, the European Space Agency’s operations account announced that ESA’s Aeolus spacecraft, an Earth science mission launched a year earlier, had performed a collision avoidance maneuver earlier that day to avoid a SpaceX Starlink satellite. This was, ESA said, the first such maneuver to avoid a “megaconstellation” satellite.
The maneuver was a success—Aeolus raised its orbit slightly to avoid the Starlink satellite—but the discussion was only starting. “As the number of satellites in orbit increases, due to ‘mega constellations’ such as #Starlink comprising hundreds or even thousands of satellites, today’s ‘manual’ collision avoidance process will become impossible,” ESA tweeted.
ESA provided no other details that day, electing to wait a day to issue a statement and not immediately responding to media inquiries. One news report, citing unnamed sources, claimed that ESA maneuvered its satellite when SpaceX “refused” to move its own satellite, making it seem the ESA satellite had the right of way and SpaceX was somehow intransigent. The event also appeared to be a harbinger of what will come with SpaceX and other companies launch thousands of satellites for their constellations.
Reality, of course, was more complex. SpaceX didn’t immediately comment on the ESA tweets, which took place on the Labor Day holiday in the US. The next day, a company spokesperson said that the company initially decided not to maneuver several days before the predicted close approach because of the low probability of a collision: 1 in 50,000. When the risk later grew, reaching a 1-in-1,000 threshold for taking action, “a bug in our on-call paging system prevented the Starlink operator from seeing the follow-on correspondence on this probability increase,” the company said.
“However, had the Starlink operator seen the correspondence, we would have coordinated with ESA to determine best approach with their continuing with their maneuver or our performing a maneuver,” the company spokesperson added.
The Starlink satellite in question, variously identified as Starlink AV and Starlink 44, was also in an unusual orbit. The satellite is one of two that SpaceX is deliberately lowering the orbits of since their launch in May as a test of their propulsion systems. That brought the satellite, intended to operate at a 550-kilometer orbit, down to around 310 kilometers, bringing it into the vicinity of Aeolus, which is in that low orbit to carry out its atmospheric science mission.
Holger Krag, director of ESA’s Space Safety Programme Office, said this was the first time controllers had maneuvered Aeolus to avoid a close approach with any space object, be it another satellite or debris, because of that low orbit. “A conjunction with an operated spacecraft is quite rare, in particular in this altitude,” he said.
All’s well that ends well, right? Well, not exactly. The close approach, and the miscommunication, illustrated the fact that the old ways of handling what’s become to be known as space traffic management won’t scale in an era when many more satellites operate in low Earth orbit.
“The case just showed that, in the absence of traffic rules and communication protocols, collision avoidance has to rely on the pragmatism of the involved operators,” Krag said. “This is done today by exchange of emails. Such a process is not viable any longer with the increase of space traffic.”
That was also the message during the Advanced Maui Optical and Space Surveillance Technologies, or AMOS, conference last week. The three-day conference, which brought nearly 1,000 people from government, industry, and academia to Maui to discuss technical and policy issues associated with space situational awareness (SSA) and space traffic management (STM), also weighed in on the incident and what it means for safe space operations.
“It made us ask whether emails or late-night calls are the most efficient coordination mechanism, and would be advisable in a scenario with thousands more operational satellites,” said Francesca Letizia, an engineer in ESA’s Space Debris Office, in one talk at the conference.
A panel that followed her speech included David Goldstein, director of special programs at SpaceX. He acknowledged that the incident was due to several factors, including the computer glitch and the long holiday weekend. “We learned a lot from that,” he said.
He also agreed that the manual coordination used today won’t work in the future. “We agree with ESA’s conclusions that it’s very difficult to be able to do this on an individual email basis,” he said. “The personal relationships are super important, but we have to figure out some automated ways to do that.”
One approach SpaceX is using is an automated system whereby the Starlink satellites themselves can maneuver automatically, based on information uploaded to them. “We built the smarts into them to be able to make some good decisions” based on that data, which includes the satellites’ own knowledge of their positions. Those decisions are doublechecked by controllers on the ground. “So far the satellites have made exquisite decisions with the information they have.” (That system, he said, was disabled on the Starlink satellite that approached Aeolus, since the company currently only uses it on satellites in operational orbits.)
ESA also is looking to automation to help solve the problem. In its original announcement about the collision avoidance maneuver, ESA said it is interested in working on approaches like the use of artificial intelligence. That work, the agency said, is dependent on funding decisions that will be made in November at the triennial meeting of the ministers of ESA’s member states, an event ESA calls “Space19+”.
Some work will take place before then. Letizia said ESA plans to start a global competition in the near future, she said, giving researchers access to historical data from ESA missions on conjunctions for analysis, allowing them to apply machine learning techniques to better understand when maneuvers are warranted.
The close approach also raises the question of, when two satellites approach, which could maneuver. Obviously if only one has the ability to maneuver, it will do so, but if both can move, the decision of which should move—expending propellant that reduces its orbital lifetime—is a more complex one.
Today there are no clear rules of the road. “The question of who moves over time will hammer itself out,” said Steph Earle, space traffic program lead at the FAA. “We’ve seen this in all modes of transportation.”
He said that, in cases like this, where both satellites are maneuverable, it’s best to let the two operators work out who should move, in order to find the most efficient approach, rather than impose a strict rule. “Is there a way to mandate who moves without considering that efficiency? I don’t think we’re there yet.”
Those efforts are only as good, though, as the data on the positions of satellites and other objects in LEO. At AMOS, many in government and industry called for greater sharing of SSA data among governments and companies. That includes the data satellite operators have on the positions of their own satellites, information that can be more accurate than from other sources.
“I think increased transparency is key. Share your data,” said Walter Everetts, vice president of satellite operations and ground development at Iridium. “There is no such thing as keeping it close the vest because that doesn’t help anybody.”
“At Planet we believe that transparency is everything, really. You cannot really hide in space,” said Agnieszka Lukaszczyk, senior director for European affairs at Planet, which operates more than 150 Earth observation satellites. “We don’t need to know what you satellite is doing. We just need to know that it’s there.”
Space Policy Directive 3, signed in June 2018 by President Trump, directs the Department of Commerce to take over civil STM services, work currently carried out by the Defense Department. That transition will take place over the next several years, but the department is working now on an “open architecture data repository” that will combine data from the DOD with that from other sources, including commercial systems.
That repository will also be a “marketplace” for new SSA providers to demonstrate their capabilities and provide value-added services on top of the basic space traffic management services that the government will continue to provide for free, said Kevin O’Connell, director of the Office of Space Commerce in the Commerce Department, at AMOS.
One issue will be validating the data and determining which sources are reliable. “Over time in a marketplace, you’ll get some sense who is better for which purposes, and who many not actually be delivering that they claim to be delivering. It sort of sorts out in the market,” he said.
Ultimately, that data could be combined with SSA data from other nations. Mark Mulholland, chief engineer for SSA and space traffic management in the Office of Space Commerce, compared SSA data sharing to the routine, free exchange of weather data among governments. That data is used in a variety of different weather models, and while governments offer free weather forecasting services based on those data and models, companies also offer value-added services.
“Isn’t this exactly what we’re trying to do in the SSA and STM enterprise?” he asked. “This is our one shot to get it right.”
The issue is a key one, although not necessarily because of the emergence of megaconstellations like Starlink. At another AMOS presentation, Darren McKnight, technical director at Centauri, argued that a bigger risk for orbital debris comes from the hundreds of large upper stages left behind in low Earth orbit, grouped in three “clusters” at 775, 850, and 975 kilometers. Two stages in the 850-kilometer cluster nearly collided in May, passing within 87 meters of each other. “If they collide, it would have doubled the catalog population in one event,” he said.
“Constellations are not bad,” he concluded. “Clusters are worse.”
Note: we are temporarily moderating all comments subcommitted to deal with a surge in spam.