Space debris includes dead satellites, abandoned rocket stages and fragments produced by explosions and collisions. Photo: iStock

On November 5, space start-up CosmoServe Space opened Base 1 in Hyderabad to tackle a problem hundreds of kilometres above us; that of cleaning up space debris. Founded by ex-ISRO engineer Chiranjeevi Phanindra, the company brought together laboratories for orbital rendezvous, robotic capture, satellite integration and mission control. The facility represents an encouraging development, but challenges include storage and final disposal.


Click the Play button to hear this message in audio format

On November 5, space start-up CosmoServe Space opened Base 1 at Gachibowli, Hyderabad, to tackle a problem hundreds of kilometres above us — the removal of space debris. Founded by Dr Chiranjeevi Phanindra, a former Indian Space Research Organisation (ISRO) engineer, the company brought together laboratories for orbital rendezvous, robotic capture, satellite integration and mission control. Union Minister Dr Jitendra Singh joined the inauguration virtually. The facility represents an encouraging development: alongside putting things into orbit, India is building the capability to remove dead satellites and other debris from orbit.

The government announcement highlighted CosmoServe’s July demonstration of soft robotic capture of debris from orbit under its Mission EMBRACE. According to a company statement, the soft robotic capture mechanism entails the use of “a compliant capture technology purpose-built to gently latch onto unprepared and non-cooperative objects in orbit — precisely the conditions found around dead satellites”. Explaining how the system works in an interview with The Federal earlier this month, Phanindra said, “We drew direct inspiration from the carnivorous Venus flytrap, whose lobes snap shut when an insect steps inside. Our soft robotic capture mechanism follows that same principle, which is why we call our capture arms ‘petals’.”

The company plans an orbital logistics hub supporting smaller autonomous vehicles for inspection, repair, refuelling and removal. Commercial operations are targeted for 2028. These remain development plans, aimed at making orbital maintenance a repeatable service.

Space debris includes dead satellites, abandoned rocket stages and fragments produced by explosions and collisions. According to the European Space Agency’s (ESA) statistics updated on 2 November, more than 47,000 objects are regularly tracked. Its model estimates about 1.5 million debris pieces between one and ten centimetres, and 230 million between one millimetre and one centimetre.

The danger is cumulative. A collision can turn two large objects into many fragments, creating fresh opportunities for further collisions. This self-generating process, associated with the Kessler syndrome (NASA space debris expert Don Kessler observed that the total amount of space debris will keep increasing; collisions will give rise to more debris, which in turn will cause more collisions), threatens useful orbital regions. ESA’s 2026 environment report warns that preventing new debris alone will be insufficient: existing debris must also be removed. What is at stake includes communications, weather forecasting, navigation and the safety of human spaceflight.

Also read: Why NASA’s Voyager 1 spacecraft reaching one light-day distance from Earth is a milestone moment

International efforts are advancing cautiously. Japan’s ADRAS-J mission inspected an abandoned rocket stage without removing it; its successor, ADRAS-J2, is intended to capture and deorbit that stage. Europe’s ClearSpace-1 mission targets the removal of the defunct Proba-1 satellite, while Britain has committed funding to remove two retired satellites. These removal missions remain planned operations.

Earlier, China’s Shijian-21 mission in 2022 moved a defunct BeiDou satellite into a higher graveyard orbit, clearing the operational region without returning it to Earth. NASA is studying the economics of remedies, including lasers that alter debris trajectories. Together, these efforts show progress, but routine orbital cleanup is still being developed.

Elon Musk has suggested that his company SpaceX’s Starship spacecraft could collect debris and in January this year expressed confidence that future capabilities would make the problem manageable. These are broad proposals. Carrying capacity alone cannot solve the difficulty of reaching and stabilising separate objects in different orbits.

SpaceX’s Stargaze system offers a more concrete contribution. It uses star trackers aboard Starlink satellites to observe objects and improve collision warnings, with conjunction data offered free to operators. It does not track every catalogued object. This supports collision prevention rather than physical removal.

Artificial Intelligence can help. AI’s potential becomes clearest during the final approach to a dead satellite. A working spacecraft can communicate its position and orientation; an abandoned one may be tumbling, reflecting sunlight unpredictably and offering no recognisable docking markers. A removal vehicle must estimate where the object is, how it is rotating and where contact can safely occur.

Machine vision can help interpret camera images and estimate a target’s relative position and orientation. NASA’s software catalogue already includes a neural-network tool for spacecraft pose estimation with applications in active debris removal. Researchers also use synthetic images to train these systems under difficult lighting, although performance must be verified against real conditions.

Beyond perception, AI-assisted planning could help select approach paths and capture opportunities, while optimisation tools could prioritise targets by collision risk, mass, orbit and removal cost. Research has demonstrated reinforcement-learning control for a debris-capture spacecraft in high-fidelity simulation. That is promising evidence of a possible capability, while flight performance still requires validation.

According to the European Space Agency’s (ESA) statistics updated on 2 November, more than 47,000 objects are regularly tracked. Its model estimates about 1.5 million debris pieces between one and ten centimetres, and 230 million between one millimetre and one centimetre. Photo: iStock

India is developing applications on both sides of the problem. CosmoServe’s founder says its capture systems use sensor fusion and machine learning to recognise objects under changing lighting and reflections, trained on simulated and laboratory imagery. This is technology under development for autonomous removal, rather than an established AI cleanup service.

The company’s Base 1 facility in Hyderabad is said to be capable of simulating space vision and microgravity to help plan and prepare retrievals from orbit.

Explaining how it works to The Federal, Phanindra had said, “Targets are identified using ground-station catalogue data for objects larger than 10 centimetres, but those coordinates can be off by up to a kilometre. A chaser can use catalogue data to get within a kilometre, but then it must rely on its own navigation sensors for the final approach and capture.” He added: “We chose vision-based optical cameras for close-range navigation. Onboard AI models, trained on synthetic images, calculate relative position and orientation down to centimetres. Space optics are challenging, though, given harsh lighting, extreme solar angles and shifting backgrounds between deep space and Earth. Our orbital rendezvous lab is a dark room with a high-intensity solar simulator and realistic Earth background projections that replicate orbital lighting. To simulate microgravity, two six-degree-of-freedom (6-DOF) robotic arms manoeuvre the chaser and target, actively counterbalancing gravity on both. When they make contact, the interaction reflects pure contact dynamics. The chaser also moves along an 8-metre linear track, creating a 13-DOF microgravity and vision testing platform.”

For tracking and cataloguing, Bengaluru-based Digantara, a space surveillance and intelligence company, offers a complementary approach. Its AIRA platform uses an AI and machine-learning processing engine to combine orbital observations from different sensors. The company has also unveiled MOSAIC, an AI-powered optical sensing network. Such systems support detecting objects, maintaining tracks and predicting their motion: the information needed to build and update an orbital catalogue, rather than physically collect debris.

The Indian National Space Promotion and Authorization Centre (IN-SPACE), an autonomous, nodal agency under the Department of Space, in its 2026 AI INSPIRED selections included Digantara’s object-evaluation software, Hyderabad-based EON Space Labs’ edge-AI payload for debris detection, and Chennai-headquartered OrbitAID’s machine-learning-assisted estimation of satellite position and orientation for servicing and refuelling. These selected development projects show an Indian pipeline from surveillance to servicing; selection does not establish successful orbital debris removal.

Collection is not like sweeping a road. A cleanup craft must match an object’s orbit and motion before contact; changing orbital planes consumes substantial fuel. Tumbling, missing capture fixtures and fragile structures complicate the grip. Contact transfers momentum, so the robot must stabilise the combined spacecraft without breaking its target. Large debris generally needs individual handling; tiny fragments require other approaches.

In his interview with The Federal, Phanindra, while explaining how space debris removal works, had said, “Anything travelling in low Earth orbit (LEO) moves at approximately 7.5 kilometres per second, or around 28,000 kilometres per hour. At that speed, even a tiny 1 or 2 millimetre aluminium ball carries kinetic energy comparable to a Formula 1 racing car hitting a stationary wall at top speed. Because kinetic energy scales with the square of velocity, speed matters far more than mass, which makes orbital impacts so catastrophic.” He added: “About 650 collisions or anomalous breakups have already generated massive debris fields. Around 50,000 objects are tracked in orbit, but only about 16,000 are operational satellites; the rest is junk. When junk strikes a working satellite, it creates thousands of fragments, each one a bullet, in a runaway cascade. If that happens, space becomes unusable. What is the point of building powerful rockets and exploring other planets if we cannot cross our own orbit because it is choked with junk?”

But removal, or collection, of the debris is not the only challenge. Storage presents another difficulty. Gathering objects in an orbital depot would require secure attachments, collision protection and continuing control. Simply parking them together creates a concentration of risk. Recycling is being studied, but routine orbital scrap processing remains a future possibility. Many removal concepts instead attach to one object and guide it towards disposal.

Where does it go? In low Earth orbit, the usual objective is atmospheric re-entry. A servicing craft can lower the target’s orbit so drag eventually brings it down, or direct a controlled re-entry towards a remote ocean region. Much material burns up, but heat-resistant components can survive; disposal planning must account for the resulting impact footprint. Returning debris intact for recovery would require a protected return vehicle and a safe landing system, adding complexity and cost.

For distant geostationary satellites, moving to a higher graveyard orbit is generally more practical than returning to Earth. The object remains in space, away from the operational belt; this relocates the hazard rather than destroying the waste. Re-entry has environmental consequences too: research has detected spacecraft-derived metals in stratospheric aerosols, while their wider effects remain uncertain. Cleanup must therefore consider both orbital safety and Earth’s environment.

Recovery also calls for attention to what remains in orbit. A malfunctioning satellite may become a long-term management problem, although a failed launch does not automatically leave persistent orbital debris. Launch reliability, satellite health and safe disposal deserve attention together.

Also read: How ISRO’s AstroSat space observatory has transformed our understanding of the universe in 11 years

India’s response extends beyond a single startup. ISRO’s Debris Free Space Missions initiative aims for all Indian governmental and non-governmental space actors to undertake debris-free missions by 2030. That is a commitment concerning Indian missions, not a promise to eliminate every fragment surrounding Earth.

The next step should be to make responsibility part of every mission’s design and budget: reserve fuel for disposal, provide capture interfaces where feasible, share orbital information and fund the removal of dangerous legacy objects. Governments can purchase early services while companies develop repeatable business models. Success should be measured by verified reductions in risk, not simply by the number of cleanup announcements.

For me, the question behind ‘The Dirty Sky’ remains urgent: having polluted the Earth, are we now polluting space? Hyderabad’s new facility offers an encouraging response. Human ingenuity is beginning to address the waste that earlier ingenuity left behind. The task ahead is to turn isolated demonstrations into dependable services, and individual national efforts into sustained international cooperation.

Next Story