How Do Satellites Avoid Collisions?

The Orbit Around Earth is Crowded

Space, often imagined as a vast, black emptiness, is surprisingly crowded. In the relatively narrow bands of orbit around the Earth, thousands of active satellites are at work. They provide global positioning, weather forecasts, internet communications, and scientific data. Alongside these operational satellites are tens of thousands of tracked pieces of “space junk,” or orbital debris. This collection includes everything from defunct satellites and spent rocket stages to tiny fragments from past collisions and tests.

This orbital environment presents a constant, high-stakes challenge: avoiding collisions. A crash in space is not a minor fender-bender. Objects in orbit travel at immense speeds – upward of 17,000 miles per hour (or 7.8 kilometers per second) in Low Earth Orbit. At these velocities, a collision with an object as small as a paint fleck can cause significant damage. A collision with a bolt or a small cube-satellite could be catastrophic, obliterating a billion-dollar spacecraft and instantly creating thousands of new pieces of lethal debris.

Managing this risk is a complex, non-stop process of tracking, prediction, and maneuvering. It’s a global effort that combines military surveillance, civilian science, and commercial innovation, all focused on a single goal: keeping satellites safe and operational.

The Crowded Sky: A New Reality for Space

The region most densely populated is Low Earth Orbit (LEO), an area extending from about 100 miles (160 km) to 1,200 miles (2,000 km) in altitude. This is where the International Space Station (ISS) and large constellations of communication satellites, like Starlink, operate. Higher up, in Geostationary Orbit (GEO) at 22,236 miles (35,786 km), satellites match the Earth’s rotation, appearing stationary in the sky. This valuable slot is also crowded with weather and communications satellites.

The space debris problem is not new, but it has accelerated. Decades of launches since Sputnik 1 in 1957 have left a legacy. Every mission that fails, every rocket body left to tumble, and every satellite that runs out of fuel becomes a long-term hazard.

Two specific events dramatically worsened the situation. In 2007, a Chinese anti-satellite weapon testintentionally destroyed a defunct weather satellite, creating a cloud of over 3,000 trackable pieces of debris. In 2009, a defunct Russian Kosmos satellite accidentally slammed into an operational Iridium communications satellite. This 2009 satellite collision was the first-ever hypervelocity crash between two intact satellites, and it instantly generated about 2,000 new pieces of large debris.

This debris doesn’t just stay in one place; it spreads out, lapping the Earth in different orbital paths and creating a minefield for every other object. This leads to the worrying scenario known as the Kessler syndrome, proposed by NASA scientist Donald J. Kessler in 1978. He theorized that if the density of objects in LEO becomes high enough, a single collision could create a cascade. The debris from one crash would hit other satellites, creating more debris, which would then hit more satellites, starting a chain reaction that could render certain orbits unusable for generations.

The Global Watchtower: Tracking Objects in Orbit

You can’t avoid what you can’t see. The foundation of all collision avoidance is Space Situational Awareness (SSA) – the discipline of tracking and understanding what is in orbit.

For decades, the primary source of this data has been the United States Space Force (USSF) and its Space Surveillance Network (SSN). The USSF’s 18th Space Defense Squadron (18 SDS) maintains a public catalog of orbital objects, sharing data with satellite operators around the world.

This network uses a variety of powerful sensors on the ground and in space.

• Ground-Based Radar: These are the workhorses for LEO. Large radar installations, like the Space Fence on Kwajalein Atoll, send out powerful radio waves. When these waves hit an object, they bounce back, allowing operators to determine the object’s altitude, trajectory, and speed. Phased-array radars can scan huge sections of the sky electronically, tracking thousands of objects simultaneously.
• Optical Telescopes: Radar is less effective for objects in very high orbits like GEO. For these, operators rely on ground-based optical telescopes. These instruments scan the sky, looking for faint pinpricks of sunlight reflecting off distant satellites. They work best at dawn and dusk, when the telescope is in darkness, but the satellite is high enough to be lit by the sun.
• Space-Based Sensors: The best view of space is from space. Satellites dedicated to SSA orbit the Earth, monitoring the environment from above, free from atmospheric distortion and bad weather that can blind ground-based sensors.

This network is incredibly capable, but it has limits. In LEO, the SSN can reliably track objects roughly the size of a softball (about 10 centimeters). In GEO, it’s limited to objects about the size of a meter.

This creates a dangerous gap. An object the size of a marble – too small to be tracked – is still large enough to destroy a satellite upon impact due to its hypervelocity speed. Satellites can be “shielded” against impacts from paint flecks and dust, but they are completely vulnerable to this lethal, untrackable debris. Collision avoidance, as it exists today, can only protect against the trackable threats.

The Process of Collision Avoidance

When the 18 SDS or a commercial tracking company identifies a potential close approach, or “conjunction,” between two objects, it sets a detailed process in motion.

Step 1: Conjunction Assessment and Data Sharing

The process begins with Conjunction Assessment (CA). High-speed computers constantly compare the predicted orbital paths of all tracked objects. When the paths of two objects are predicted to come uncomfortably close, an alert is generated.

This alert, often called a Conjunction Data Message (CDM), is sent to the operators of the active satellites involved. A CDM contains information:

• Which two objects are involved (one is often an active satellite, the other debris or another satellite).
• The time of closest approach (TCA).
• The predicted miss distance, broken down into radial (height), in-track (along the orbit path), and cross-track (sideways) components.
• The Probability of Collision (Pc).

No orbital prediction is perfect. An object’s future position is always a matter of probability, not certainty. This is because forces like atmospheric drag (which changes with solar weather), the pull of the sun and moon, and the Earth’s uneven gravity all introduce tiny errors that grow over time.

This uncertainty is visualized as a “pizza box” – a 3D error bubble around the satellite’s predicted location. The collision risk isn’t just about the centers of two satellites getting close; it’s about whether their “pizza boxes” overlap.

This is where data sharing becomes essential. Commercial operators like Starlink and OneWeb share their orbital data and maneuver plans. Organizations like the Space Data Association (SDA) act as a clearinghouse, allowing operators to coordinate directly, a process often called “operator-to-operator” communication. This ensures one operator doesn’t move their satellite directly into the path of another.

Step 2: Risk Analysis

An operator might receive dozens of these alerts every day for a single satellite. Most are not serious. The initial warnings can come 72 hours or more before the potential conjunction. As the TCA gets closer, ground stations get more tracking data, the orbits are refined, and the “pizza boxes” shrink. In most cases, the new data shows that the objects will miss each other by a safe margin, and the alert is cleared.

But sometimes, the risk remains high. Operators must decide on an acceptable level of risk. For most robotic satellites, a common threshold is a probability of collision of 1 in 10,000. If the CDM shows the risk is higher than this (e.g., 1 in 5,000), the operator will plan a maneuver.

This 1-in-10,000 number is not arbitrary. It’s a careful balance. If the threshold is too low (e.g., 1 in 1,000,000), operators would be maneuvering constantly, wasting fuel for alerts that were likely false alarms. This is the “crying wolf” problem. If the threshold is too high (e.g., 1 in 100), it invites disaster. Human-rated spacecraft, like the ISS, use a much stricter threshold, often 1 in 100,000.

Step 3: The Maneuver Decision

The “go/no-go” decision is usually made between 8 and 24 hours before the TCA. This is a tense calculation for the satellite operator.

First, a maneuver costs fuel. Most satellites carry a finite amount of propellant. This fuel is needed for station-keeping (staying in the correct orbit) and for the satellite’s eventual, safe de-orbit. Every collision avoidance maneuver shortens the satellite’s operational lifespan, which can cost a company millions in lost revenue.

Second, a maneuver disrupts the mission. A satellite observing the Earth might have to stop imaging. A communications satellite might have to hand off its data traffic. These service interruptions are undesirable.

Third, the maneuver itself carries risk. Firing a thruster is a complex event. There’s also the chance that the maneuver, while dodging one piece of debris, could accidentally move the satellite into the path of a different piece of debris that wasn’t part of the original calculation. Operators must run new screenings on the planned post-maneuver orbit to ensure it’s also clear.

Step 4: Executing the Avoidance Maneuver

If the decision is “go,” the operator’s flight dynamics team calculates the burn. The goal is to be as efficient as possible. They aren’t trying to swerve violently out of the way. They just need to ensure the satellite and the debris don’t arrive at the same point in space at the same time.

A typical maneuver is a very small nudge. By firing its thrusters for just a few seconds, the satellite can slightly change its speed.

• Speeding up (a prograde burn): This raises the satellite’s orbit, making it take longer to circle the Earth.
• Slowing down (a retrograde burn): This lowers the satellite’s orbit, causing it to complete its orbit faster.

This small change in-tracking is all that’s needed. By the time the satellite reaches the conjunction point, it might be 30 seconds earlier or later than originally predicted. At 17,000 miles per hour, 30 seconds of timing difference translates to a miss distance of more than 100 miles.

Maneuvers that change the tilt of the orbit (cross-track burns) are much less common. They are “energetically expensive,” meaning they consume a very large amount of propellant, so operators avoid them unless absolutely necessary.

After the maneuver, ground stations track the satellite to confirm its new orbit is correct and the threat has passed.

The Rise of the Megaconstellations

This manual, human-in-the-loop process works for a few hundred or even a couple of thousand satellites. It breaks down completely at the scale of modern “megaconstellations.”

Companies like SpaceX and OneWeb are building constellations that involve tens of thousands of satellites. A single operator can’t manually analyze thousands of conjunction alerts every day. The new reality has forced a move to autonomous collision avoidance.

Satellites in these large constellations are designed to dodge threats on their own. They receive conjunction data from the U.S. Space Force and commercial trackers like LeoLabs. On-board flight computers analyze the risk. If a threat meets the pre-programmed risk threshold, the satellite’s software automatically calculates and executes a maneuver, all without a human operator intervening.

Starlink’s system is the most prominent example. Their satellites use electric hall thrusters to make very small, efficient adjustments. SpaceX has reported its system performs thousands of automated maneuvers every month.

This automation creates new challenges. The biggest is the “right of way.” If two active satellites from different companies are on a collision course, who moves? Both are maneuverable. Both have automated systems. This has led to a call for internationally recognized “rules of the road” for space. Right now, deconfliction often happens through emails and phone calls between operators, but this won’t be sustainable as constellations grow.

Protecting the International Space Station

The International Space Station is a special case. As a crewed, bus-sized object in a debris-filled orbit, it requires the highest level of protection.

NASA’s Johnson Space Center in Houston is responsible for ISS safety. They work with the 18 SDS and Roscosmos to monitor all threats. The ISS is heavily shielded, but it’s not invincible.

If a piece of debris is projected to pass through a “pizza box” around the station that is roughly 25 miles wide, 25 miles long, and 0.5 miles high, a “Debris Avoidance Maneuver” (DAM) is planned. These maneuvers are a big deal. They involve firing the thrusters on the station’s Zvezda module or on a docked resupply ship, like a Progress or Cygnus vehicle. The entire half-million-kilogram station is gently pushed into a slightly higher orbit.

This has happened dozens of times over the station’s history.

Sometimes, a threat is spotted too late. If a high-risk conjunction is detected with too little time to perform a DAM, flight controllers will order the astronauts to “shelter in place.” This involves the crew closing hatches between the station’s modules and moving into their docked “lifeboat” spacecraft, such as the Crew Dragon or Starliner. They wait there during the time of closest approach. If the station were to be hit and depressurize, the crew would be safe inside their capsules, ready to undock and return to Earth if necessary.

Long-Term Solutions: Cleaning Up Space

Collision avoidance is a reactive, short-term fix. It treats the symptoms, not the disease. To ensure the long-term sustainability of space, two other strategies are being pursued: mitigation and remediation.

Mitigation Guidelines

Mitigation means not making the problem worse. The primary international guideline, adopted by space-faring nations, is the “25-year rule.” This is a commitment that all new satellites launched into LEO must be able to remove themselves from orbit within 25 years of their mission ending.

This is usually done in one of two ways. The satellite can save just enough fuel to perform a final “de-orbit burn,” actively slowing itself down so it re-enters and burns up in the atmosphere. Or, if it’s in a high-enough orbit, it can be lowered to an altitude where atmospheric drag will naturally pull it down within 25 years.

Another key mitigation rule is “passivation.” This involves safely venting any leftover propellant and discharging batteries on a satellite at the end of its life. This prevents the satellite from exploding due to overheating or pressure buildup, an event that can create clouds of new debris.

Active Debris Removal (ADR)

Mitigation only applies to new satellites. It does nothing about the thousands of defunct satellites and rocket bodies already floating in orbit. For that, the world is slowly developing Active Debris Removal (ADR).

ADR is the concept of a “space garbage truck.” A specialized spacecraft would be launched to rendezvous with a large, dangerous piece of debris, capture it, and then drag it down to burn up in the atmosphere.

Several companies and agencies are testing this technology.

• Astroscale: This Japanese company successfully tested its ELSA-d mission, which used a magnetic system to capture a test satellite.
• ClearSpace: This Swiss startup was commissioned by the European Space Agency (ESA) for a mission to capture a large piece of a Vega rocket with a large, four-armed robotic claw.

The challenges for ADR are immense. It’s technically very difficult to rendezvous with and grab an object that is tumbling uncontrollably. It’s also expensive, with each removal mission costing tens or hundreds of millions of dollars. Finally, it’s legally and politically complex. Does a company from one country have the right to grab another country’s (defunct) satellite? These are a few of the questions being worked out.

The Future of Space Traffic Management

The current system – a military-run catalog, informal data sharing, and operator-by-operator collision avoidance – is being stretched to its limit. The consensus is that the world needs a proper Space Traffic Management (STM) system.

This would be the space-based equivalent of Air Traffic Control (ATC). In the United States, this responsibility is being shifted from the Department of Defense to a civil agency, the Department of Commerce’s Office of Space Commerce.

A future STM system would provide a single, open, and more accurate catalog of space objects, incorporating data from all public and private sensors. It would establish clear “rules of the road” for right-of-way and maneuvers. It would also use Artificial Intelligence to predict risks with greater accuracy and automate deconfliction between operators.

Summary

Avoiding collisions in space is a dynamic and relentless task. It relies on a global network of sensors to track tens of thousands of objects, from active satellites to hazardous debris. The process involves a careful analysis of risk, balancing the probability of a catastrophic impact against the high cost of maneuvering.

As Low Earth Orbit becomes increasingly crowded with new megaconstellations, this system is undergoing a rapid evolution. The manual, human-steered process is giving way to automated, AI-driven collision avoidance systems. Looking forward, the long-term health of the orbital environment will depend not just on dodging debris but on mitigating the creation of new junk and, perhaps, actively removing the most dangerous pieces already there. Keeping space safe is essential for preserving the modern technologies that depend on it.