Autonomous Swarm Robotics for Space Debris Removal — ELSA-M and ClearSpace-1 Herald a New Commercial Market

Every satellite in low Earth orbit travels at roughly 7.8 kilometers per second. At that speed, a bolt the size of a fingernail carries the kinetic energy of a hand grenade. There are 36,500 catalogued objects larger than 10 centimeters up there and no one is coming to pick them up. Except, starting this year, a small group of companies that have spent the last decade proving it can be done.

The math is straightforward: every year the orbital population grows faster than natural decay removes it. NASA's Orbital Debris Program Office estimates that even with zero new launches, the existing debris field would keep generating fragments through cascade collisions. Active debris removal (ADR) is no longer a hypothetical insurance policy. It is becoming an operational requirement.

The long road from concept to first capture

The idea of removing defunct satellites predates the commercial space age, but practical demonstrations only began in earnest during the last decade. In 2018, the RemoveDEBRIS mission (a University of Surrey-led project deployed from the ISS) tested a net and a harpoon against artificial targets. It worked. The net wrapped around a mock satellite and the harpoon pierced a panel. Both methods proved that capture in microgravity was mechanically feasible.

Astroscale followed with ELSA-d in 2021, the first demonstration of magnetic docking plate technology. A servicer spacecraft released and recaptured a simulated debris object repeatedly, proving that rendezvous and proximity operations could be executed autonomously with sub-meter precision. The company followed up with ADRAS-J in 2024, an inspection mission that approached a defunct Japanese rocket stage to within meters and mapped its condition and tumbling state. That data is critical for planning a future removal.

ClearSpace-1, the European Space Agency's flagship ADR mission, targets a different challenge. Its target, PROBA-1 (an ESA satellite launched in 2001), was never designed for removal. It uses a four-armed claw mechanism called the Claw to grapple the tumbling object and stabilize it before deorbit. The mission was originally scheduled for 2026 but is now planned for 2028. The claw capture system has completed ground-based hardware-in-the-loop validation, and the company is actively hiring robotics engineers for the Luxembourg assembly and integration facility.

TIMELINE: Autonomous Debris Removal
─────────────────────────────────────────────────────────────
  2018 ──── 2021 ──── 2024 ──── 2026 ──── 2028+
  🎯       🧲       🔭       🚀       🔮
  Remove-  ELSA-d   ADRAS-J  ELSA-M   Clear-
  DEBRIS   (mag     (inspec-  (com-    Space-1
  (net +    dock)    tion)     mercial  (unpre-
   harpoon)                    demo)    pared)
─────────────────────────────────────────────────────────────
Sources: NASA, ESA, Astroscale, ClearSpace

Two companies, two technological paths

The current ADR field splits into two primary architectural approaches. Astroscale's ELSA family uses magnetic docking plates that must be installed on satellites before launch. The servicer spacecraft approaches, aligns with the plate, and uses magnetic attraction to establish a rigid capture interface. Once secured, the servicer fires its thrusters to deorbit the combined stack. The advantage is simplicity and repeatability: the same servicer can service multiple targets in sequence if they share the docking plate standard.

ClearSpace's Claw system takes the opposite approach: it targets unprepared objects. The four-armed mechanism can envelop irregular shapes, adjust for rotational motion in real time, and apply controlled force without damaging fragile structures. The trade-off is complexity: the claw requires multi-axis force sensing, vision-based pose estimation, and autonomous grasp planning. As we wrote in July, space-grade AI accelerators like EdgeCortix's SAKURA-II are now reaching orbit with radiation-hardened neural network inference capabilities that make this level of onboard autonomy feasible.

Starfish Space occupies a middle ground. Its Otter servicing vehicle is designed for cooperative docking but can also perform debris disposal missions. The company secured a $29 million Series B in November 2024, bringing total funding past $50 million, and has signed contracts with Intelsat and the U.S. Space Force. The Otter Pup demonstration mission flew in 2023 and validated the core rendezvous and capture software stack.

The swarm question: from single capture to multi-agent operations

Every ADR mission flown or funded to date is a single servicer targeting a single object. The next logical step, and the one that could transform the economics of debris removal, is to scale from bespoke salvage operations to fleet-based services using cooperative multi-agent systems.

Swarm robotics research for space applications has accelerated in the past two years. A 2026 detailed review in ScienceDirect catalogued over 80 papers on swarm intelligence for space robotics, covering applications from on-orbit servicing to planetary exploration.

PitchBook's 2026 Emerging Space Brief on swarm robotics notes that the technology is at the boundary between academic research and commercial application. The Defense Advanced Research Projects Agency (DARPA) has funded multi-agent rendezvous and proximity operations through its orbital logistics programs, and SpaceX's Starship-based deployment capability could, in principle, deliver a constellation of small servicer spacecraft to LEO in a single launch, which would fundamentally change the unit economics of debris removal.

The fundamental challenge is not mechanical. Multiple companies have proven that individual capture works. The open question is coordination: can a decentralized group of autonomous servicers negotiate capture assignments, avoid collisions with each other and with operational satellites, and sequence deorbits to minimize fragmentation risk, all without a human in the loop?

The bottleneck shifts to regulation

As ADR technology matures from TRL 5 to TRL 7, the binding constraint is no longer engineering. It is the legal and regulatory framework for removing another party's property from orbit. Current international space law, anchored in the Outer Space Treaty of 1967 and the Liability Convention of 1972, was written when the concept of a commercial salvage vehicle was science fiction. There is no established protocol for obtaining consent to remove a defunct satellite owned by a bankrupt entity or a government that no longer exists.

Astroscale's ELSA-M solves this by targeting OneWeb satellites equipped with pre-installed docking plates. The owner has consented and the interface exists. It operates under a direct ESA contract, so ownership is unambiguous. A future commercial swarm operator removing mixed-ownership debris would need a far more complex liability framework. The U.K. Space Agency launched a tender in June 2026 for regulatory studies on active debris removal licensing, suggesting that governments are beginning to address the gap.

The FCC's 2024 orbital debris mitigation rule, requiring disposal within five years of mission completion for U.S.-licensed satellites, creates a downstream compliance driver, since operators who fail to deorbit become removal candidates. NASA's 2026 State of the Art report notes that "debris regulations are becoming more stringent" and that "compatible systems for spacecraft rendezvous and removal are being developed in parallel" with the regulatory push.

Sources