Collision Avoidance

Optical Measurement

NASA

The Orbital Debris Engineering Model (ORDEM) is NASA’s primary engineering tool for describing the orbital debris environment; that is, how many pieces of space debris are in Earth orbit, how often spacecraft encounter them, and what kinds of particles are present at different altitudes. It is developed and maintained by the NASA Orbital Debris Program Office.
ORDEM lets engineers and mission planners quantify debris flux, the rate at which debris of various sizes passes through a given orbit. It helps determine how many pieces of debris a spacecraft might run into over time and at what sizes, speeds, and directions those impacts are likely. Unlike simple debris counts, ORDEM gives engineering-useful outputs (e.g., debris flux by size and velocity) that directly feed design decisions, such as shielding thickness and risk assessments.
The ORDEM model series has been continuously updated over decades, with major releases such as ORDEM 2.0, ORDEM 3.0, and ORDEM 3.1, each incorporating more data and extending model capabilities. Newer versions (e.g., ORDEM 3.2) refine flux estimates and make the tool easier to use for current spacecraft and missions.

The Model Process document explains how ORDEM 3.1 was built from observational datasets, analytical models, and statistical techniques and outlines the physical and mathematical foundations that determine how debris populations are represented in the model.

ORDEM 3.1 supersedes the earlier ORDEM 3.0 model, replacing it with updated debris populations and advanced analytical methods.

It provides annual average estimates of debris flux (how many particles of a given size pass through a particular orbital volume in a year). This makes it suitable for long‑term design and planning, but not for predicting collisions immediately following a new debris‑generating event (e.g., satellite breakup).

ORDEM does not propagate individual debris orbits precisely like a tracking catalog; instead it offers a statistical description of debris populations based on available measurements.

The model describes the debris environment in terms of:

Debris size: from micrometer scale (≈10 µm) up to meter scale.

Orbital regions: from Low Earth Orbit (LEO) through Medium Earth Orbit (MEO) and up to Geosynchronous Orbit (GEO).

Material density categories: debris is grouped by density (e.g., low, medium, high, and sodium‑potassium droplets) to better approximate its impact risk and behavior.

The model outputs flux estimates how many debris particles of a specified size and material category pass through a unit area per year in a given orbit. These fluxes are critical for assessing collision risk and for spacecraft shielding design.

ORDEM 3.1 is based on multiple types of data sources:

1. Space Surveillance Network (SSN) Catalog

Provides data on cataloged objects in Earth orbit.

SSN covers objects down to about 10 cm in LEO and about 1 m in GEO.

These cataloged objects form the deterministic “backbone” of the model for larger debris.

2. Ground-based Radar Observations

Radar data (especially from the **Haystack Ultrawideband Satellite Imaging Radar; HUSIR) help characterize debris between the SSN threshold (≈10 cm) down to a few millimeters.

Radar datasets often require data cleaning and statistical processing to remove noise and ensure accurate population estimates.

3. In Situ Impact Data

Measurements of debris impacts on returned spacecraft surfaces (e.g., Space Shuttle windows and radiators) help characterize sub‑millimeter debris that cannot be observed by ground sensors.

These in situ records are used to extrapolate the small particle populations that pose significant spacecraft risk yet are too small to be cataloged directly.

4. Optical Observations

Optical telescopes like the Michigan Orbital Debris Survey Telescope (MODEST) help characterize debris in high orbits (e.g., GEO), particularly for objects larger than ≈10 cm.

5. Simulation and Supporting Models

ORDEM 3.1 uses LEGEND, a long‑term orbital debris evolutionary model, to provide reference population statistics, which are then scaled and adjusted using real observational data.

Additional models handle debris fragmentation, atmospheric drag effects, and simulation of future debris generation processes (explosions, collisions) using tools like the NASA Standard Satellite Breakup Model.

Debris populations are binned according to:

Orbital altitude bands

Inclination ranges

Eccentricity ranges

Size ranges

Material density categories

Because observational data are incomplete and indirect (e.g., radar cross sections do not map perfectly to physical size), ORDEM 3.1 uses Bayesian and maximum likelihood statistical techniques to:

Scale the initial reference populations (from LEGEND)

Adjust population densities where radar and in situ data provide reliable samples.

Interpolate between measured data points and extrapolate to unmeasured regions of size and orbit.

One of the key advancements of ORDEM 3.1 is its explicit consideration of uncertainties in debris flux predictions:

Uncertainties arise from measurement errors, limited data sampling, and statistical model assumptions.

ORDEM 3.1 quantifies these uncertainties so users can understand confidence bounds around debris flux estimates, a crucial factor in risk assessments.

ORDEM 3.1 interpolates debris flux across 11 fiducial size points using techniques such as Piecewise Cubic Hermite Interpolating Polynomial (PCHIP), ensuring smooth and physically reasonable flux values between measured or estimated data points. This allows the model to produce continuous flux estimates across size ranges that might have sparse observational coverage.

Flux outputs include:

Spacecraft mode: flux presented in terms of local azimuth, elevation, and relative velocity as seen by an orbiting spacecraft.

Telescope/radar mode: flux as detected from a ground‑based sensor pointing at specific angles.

These dual modes support both mission design and observational planning.

Space-X's Starlink

Starlink's collision avoidance strategy is multi-layered and includes both passive design choices and active operational processes:

Passive Orit Designs.

Station‑keeping slots:
Each Starlink satellite is assigned a specific orbital slot with defined orbital parameters (altitude, inclination, phase). This structured layout ensures that while satellites remain in their slots they are geometrically separated — minimizing risk of collision within the constellation itself (“Starlink‑on‑Starlink”).

Orbit selection:
Starlink satellites are placed in relatively low altitudes (~540–550 km). These lower orbits enhance natural atmospheric drag, which ensures defunct satellites decay and reenter the atmosphere within ~5 years, preventing long‑lived debris accumulation.

Active Collision Avoidance Systems.

1. Automated Maneurvering.

Autonomy:
Satellites ingest tracking data (primarily from the U.S. Space Force’s 18th Space Control Squadron) and automatically assess conjunction risk against known space objects. If a dangerous close approach is predicted, the satellite can perform an avoidance maneuver without real‑time human intervention. Oversight teams remain available for supervision and decision escalation if needed.

Thruster system:
Each vehicle uses krypton‑powered Hall effect thrusters, which provide efficient propulsion for both routine station‑keeping and collision avoidance burns. These thrusters enable precise velocity adjustments to alter orbits sufficiently to avoid predicted conjunctions.

2. Collision Avoidance Thresholds

Conservatism:
SpaceX uses very strict collision probability thresholds to decide when to maneuver. Starlink reportedly initiates avoidance actions when collision risk is as low as 1 in 1,000,000, which is roughly two orders of magnitude more conservative than many industry norms (often ~1 in 10,000). This conservative stance leads to frequent maneuvers but reduces risk significantly.

Frequent data updates:
Starlink satellites update orbital predictions and tracking data every ~30 minutes, allowing the system to refine conjunction assessments with relatively fresh information.

3. Coordination with Third Parties

Screening APIs: Starlink offers Space Traffic Coordination APIs, through which other satellite operators can submit their ephemerides (orbit data) for screening against Starlink’s constellation. Results are returned quickly (typically in minutes) and can be used to plan maneuvers in an integrated traffic picture.

Manual coordination: If another operator prefers to perform a maneuver themselves during a conjunction event, SpaceX’s system allows Starlink satellites to hold their orbit (remain ballistic) for the duration of the avoidance scenario if instructed by the other party. However, due to the lack of standardized automated arbitration, Starlink satellites currently default to taking maneuver responsibility for most third‑party conjunctions.

NASA cooperation: SpaceX and NASA have a Joint Spaceflight Safety Agreement under which both entities exchange precise orbital and tracking data and agree on maneuver responsibilities for conjunction events involving NASA spacecraft. Unless specifically coordinated otherwise, Starlink will maneuver to avoid NASA assets, with NASA refraining from additional maneuvers to prevent inadvertent conflicts.

Types of Collision Avoidance Scenarios:

1. Starlink-on-Starlink

Internal deconfliction: Passive spacing and station‑keeping ensure that satellites within the constellation rarely collide thanks to planned orbital geometry, this is the primary layer. The active avoidance system functions as a secondary defense to correct any anomalies or unexpected conjunctions inside the constellation geometry.

2. Starlink-on-Debris

Space debris: The satellite assesses conjunctions with debris tracked by the U.S. Space Surveillance Network and initiates avoidance if risk exceeds its threshold. Since space debris includes a vast catalog (~tens of thousands of objects ≥10 cm plus millions of smaller fragments), this accounts for the majority of avoidance maneuvers.

3. Starlink-on-Other-Operators

Other satellites: Starlink spacecraft calculate conjunction risks with satellites launched by other nations and companies. Because operators do not yet have a universal automated coordination protocol, Starlink defaults to assuming responsibility for avoidance and will maneuver unless external operators request otherwise.

Uncoordinated conjunction risks:
Recent near‑collision events (e.g., with a Chinese spacecraft) highlight the limitations: if orbit data isn’t shared, autonomous systems cannot predict close approaches reliably, raising operational risk.

4. Human Spaceflight & Government Assets

ISS and crewed vehicles: Starlink orbits are macro‑deconflicted with critical human spaceflight operations such as the International Space Station, meaning the constellation is designed to avoid appearing in ISS risk assessments and reduce unnecessary operational actions by NASA or partner agencies. Coordination is maintained with NASA when inbound or outbound crew/cargo missions require it.

Scale and Operational Reality of Maneuvers

Starlink satellites have cumulatively performed tens to hundreds of thousands of avoidance maneuvers over the past few years. Some analyses suggest hundreds of maneuvers per satellite annually, with the frequency tied to orbital congestion and conservative risk thresholds.

These operations are part of daily orbital maintenance and planning; autonomous systems, not manual intervention, handle the bulk of them.

Japan (JAXA)

A centerpiece of Japan’s collision avoidance capability is the RABBIT system:

What RABBIT Is:

Acronym: Risk Avoidance Assist Tool based on Debris Collision probability.

A software tool developed by JAXA’s SSA team to interpret conjunction data and produce maneuver planning guidance.

Functionality:

RABBIT takes Conjunction Data Messages (CDMs) issued by the U.S. Combined Space Operations Center (CSpOC) ,which list predicted close approaches and estimated risk parameters.

The tool automates collision risk computation and displays it visually (e.g., risk contours), enabling operators to determine when and how to execute debris avoidance maneuvers.

It quantifies how proposed avoidance maneuvers (e.g., small delta‑V changes) would reduce collision probability over time.

Operational Use and Availability:

RABBIT allows satellite operators, including those without deep orbital dynamics expertise, to rapidly design optimal avoidance plans based on CDM input.

JAXA provides the tool free of charge to operators worldwide to support sustainable operations and reduce collision risk industry‑wide.

Japan’s collision avoidance activities are reinforced through global partnership and data exchange:

Bilateral SSA Collaboration with the United States:
Under a formal SSA cooperation framework, JAXA and U.S. Space Force’s Combined Space Operations Center (CSpOC) exchange space‑object orbit data regarding debris and satellites.

Integration with Japanese Defense SSA:
The Japan Air and Space Self‑Defense Force’s Space Operations Group combines sensor data from Japanese radar networks with external sources (including U.S. contributions) to cultivate national space domain awareness and support civil and military space safety.

Ongoing Technological Enhancements:

Radar and optical systems enhancements to expand detection sensitivity to smaller debris and increase observational coverage.

Advanced analysis systems developed with partners like Fujitsu support automated planning, batch processing of observational inputs, and rapid maneuver recommendation generation.

Future expansion of a national SSA capability integrates JAXA and defense STA systems under Japan’s Basic Plan on Space Policy, focusing on resilient and comprehensive collision risk capability.