On May 16, a very special accident happened. About 700 km above our head, the satellite at Sentinel-1 developed an emergency strategy to avoid the wreckage that was hitting it: the remnants of a Russian satellite were deliberately destroyed by Moscow in November 2021. At the time, these anti-satellite missile tests created a small shock wave in the face of the risk of creating new hazardous debris for machines in operation.
A prophecy is briefly avoided this time. On Twitter, the European Space Agency (ESA), which is in charge of the satellite, said: “The situation developed rapidly, was complicated to avoid collisions and we had less than 24 hours to work.” Despite this difficulty, the satellite was able to avoid small debris of a few centimeters which could be fatal.
This adventure, if successful, raises some questions about the safety of satellites in orbit around the Earth. “These risks are becoming more and more frequent,” assures Jean-Paul Naib, director of space at the Federal Polytechnic School of Lausanne (EPFL). Now once a month, the International Space Station needs to strategize to avoid the worst. A complex situation that is not expected to improve as the number of satellites increases with each passing day, and no immediate solution seems to be possible to reduce this number.
AI for rescue
For some, getting rid of artificial intelligence can be a daunting task. Towards the end of 2019, ESA announced the development of a system that would allow satellites to detect their own collision risk and act accordingly. Two years later, during the European Conference on Space Debris, the agency assured that their results were positive and that in their tests, the satellites were able to make the right decision at risk. Ralph Densing, director of operations, said: “Our results are not perfect, but in many cases artificial intelligence was able to accurately detect where action was needed.” For the moment, their role is limited to sending a warning to ground teams to practice tactics, but the idea is to change them 100% automatically.
“As far as computer calculations are concerned, this is something that is within our reach,” explains Jean-Paul Kenib. But the main obstacle is at the catalog level. “If the goal is to get rid of the heavy and costly task of avoiding debris on the computer, they must provide all the necessary data. However, this information is only partial. There are many catalogs of space debris but they are not complete. Military satellites are also invisible for security reasons.
With these limitations, the fully automated approach becomes more complicated, according to Jean-Paul Kieb: “Without a complete catalog, we cannot be sure that the strategy does not lead to new risks. , Which is very difficult to set up.
Space Garbage Truck
However, time is running out. Currently, each space mission spends 5 to 10% of its budget to avoid debris, according to the OECD. And if the risk increases, the cost will also increase. These strategies require fuel and the later the risk becomes visible, the greater the change in trajectory. Not to mention the need to have a dedicated team to monitor 24 hours a day. “Ideally, an artificial intelligence should alert us to the risk of a collision one to two weeks before the injury,” said Jean-Paul Kenneth. It takes a lot of confidence. “
Another solution is to reduce the amount of waste, and even there artificial intelligence seems to be a solution. Swiss start-up Clearspace is currently developing a “space debris truck” that will have to drag debris into orbit and return it to Earth’s atmosphere where it will burn if it falls. And it’s a computer that manages the system! A complex maneuver as the debris in question turns on and off by itself. The project is promising but does not reduce the risk in the short term. “If we want to use Earth’s orbit, we must quickly find a solution,” argued Jean-Paul Kenneth. “Our satellites are not designed to constantly change orbits.”