24 tiny CubeSats turned a dead satellite in simulation

What happened

A car breaks down in a car park. Nobody can start it. So four people put their hands on it and push. One person alone would fail. Four people, pushing in step, can steer it into a space.

Dead satellites have the same problem. They still fly, but nothing on board answers. Many of them tumble. You cannot push a tumbling object out of the sky safely. First you have to stop it turning and point it the right way.

A new study tests a fix. Stick a swarm of shoebox-sized satellites onto the dead one. Fire their tiny thrusters in step. Use them as the steering the dead craft no longer has.

The test

The paper is “Control of Decommissioned Satellites and Space Debris Using CubeSats with Ion Electrospray Engines”. It was written by Felix Biertümpfel, Peter Seiler, Paulo Lozano and Harald Pfifer. They work at TU Dresden in Germany, the University of Michigan in the United States, and MIT in the United States. It went on arXiv on 31 August 2026.

Everything here happens in a computer. The team built a Matlab model of a dead satellite in a 685 km orbit. The main body weighs 500 kg. Two floppy solar wings hang off it, 42 kg each, and they bend and twist.

Onto that they stick 24 small satellites. Four on each of the six faces. Each one weighs 3.5 kg and is the size of three juice cartons stacked up. Each carries three stages of tiny thrusters, 32 per stage. One stage pushes with 0.64 thousandths of a newton. That is about the weight of a grain of sand.

Together the 24 give a turning force of up to 4.61 thousandths of a newton metre. That is at least 2.8 times the worst push the sun, the air and the Earth’s field can give the satellite. With the swarm on board, the whole thing weighs 668 kg.

The hard part is that nobody knows the dead craft well. So the team let key numbers wander. Weight could be off by 20 percent. Spin resistance by 17 percent on each axis. The solar wings could sit at any angle at all. They then built one controller that had to work across every one of those cases.

The result

In the model, the swarm turned the dead satellite halfway around, 180 degrees, in 1,228 seconds. That is about 20 minutes. The worst pointing error during the turn was 0.6 degrees, and it came at one moment when the thrusters ran out of push.

The turn used 0.00028 kg of fuel. That cost less than 0.1 metres per second from the swarm’s speed budget. The team then reran the turn 256 times, once for every worst-case mix of the unknown numbers. Tracking barely changed.

A second test held the satellite still for a full orbit while the sun and the air pushed on it. The target was 60 arcseconds, which is a sixtieth of a degree. The swarm stayed inside that with room to spare.

What it means

Old satellites are the problem, not new ones. The paper cites an ESA report saying about 40 percent of retired low orbit satellites missed the old 25 year clean-up rule. Those craft are already up there. They carry no fuel and no plan.

Big rescue spacecraft cost a lot. This study skips them. The swarm rides up on a shared launch, flies to the target on its own, sticks on, and takes over the steering. No mothership.

There is a second use. A working satellite often dies because its steering fails or its fuel runs dry. Stick the same swarm on, and the craft keeps working. The paper flags this as a way to extend life, not just to clear rubbish.

Business ideas from this paper

1. Sell a drop-in docking pad and marker kit so new small satellites can be grabbed later. Who buys it: firms that build satellites under 1000 kg and need to meet clean-up rules. A price to test: 4,000 dollars per kit, drawings included. A one-week test: draw one pad, publish the drawing free, and email 20 satellite builders asking if they would fit it.
2. Sell a sizing tool that says how many small satellites a given dead craft needs, using the torque sums in this paper. Who buys it: debris removal startups and the insurers who price their missions. A price to test: 79 dollars a month. A one-week test: build it in a spreadsheet, size three real retired satellites, and send the answers to ten removal startups.
3. Sell a ready-made simulation pack of a tumbling dead satellite for control engineers to test their own designs against. Who buys it: university control groups and space firm engineering teams. A price to test: 300 dollars a licence. A one-week test: post a cut-down free version, then count how many of the first 50 users ask for the full one.

How sure can you be?

Nothing flew. This is a computer model from start to finish. No hardware was tested and no dead satellite was touched.

The paper only covers pointing and turning. Getting the swarm to the target, closing in and sticking on are all named as hard steps and left to other work. The authors also skip the sum that shares the wanted force out among the 24 craft. They assume the swarm can read the dead craft’s attitude, either by radio or by its own star trackers.

Speed budget is thin. One thruster stage gives the joined craft only 6.8 metres per second. All three stages give about 20. Even in the best case, with the craft rotated so every thruster helps, the paper says 125. That is enough to nudge an orbit, not to drag a satellite down from anywhere.

THE PAPER ALSO ADMITS THAT SPENT THRUSTER STAGES BECOME DEBRIS THEMSELVES. It says that problem is for a later study. The work is a preprint, names no journal, and releases no code. To settle it, somebody has to fly a swarm, dock it, and turn a real dead satellite.

Do this today

If you build satellites, add a flat, marked spot that something can grip later. It costs almost nothing now and it is the one thing every removal plan in this paper depends on.

Source: Control of Decommissioned Satellites and Space Debris Using CubeSats with Ion Electrospray Engines, August 2026. arXiv:2608.30215 · arxiv.org (preprint · not yet peer reviewed).

Just Out Tech explains new research in plain language. This article was drafted with AI assistance and checked by a human against the original source.