Technology • July 25, 2025
With over 27,000 tracked debris objects and millions more too small to catalog, autonomous collision avoidance is no longer optional — it is a prerequisite for any responsible orbital operator.
"Ground-based tracking and uplinked avoidance maneuvers are a 1990s solution to a 2030s problem. Autonomous onboard decision-making is the only architecture that scales."
— PAVE Space GNC Lead
The orbital environment has changed dramatically over the past decade. The rapid proliferation of large commercial constellations, combined with a legacy population of defunct satellites and rocket bodies, has transformed low Earth orbit into one of the most operationally complex environments that any autonomous system must navigate. For an orbital logistics vehicle — one that must not only maintain its own orbit but actively maneuver to rendezvous with and service other spacecraft — collision avoidance is not a background function. It is a mission-critical capability that must operate reliably with minimal latency.
Today, PAVE Space is sharing a detailed overview of its Autonomous Debris Avoidance System (ADAS), a core subsystem of our orbital transfer vehicle platform that enables real-time threat detection, trajectory assessment, and evasive maneuver execution — entirely onboard, without dependence on ground station contact.
Current industry-standard practice relies on conjunction analysis performed on the ground using Space Surveillance Network (SSN) data, followed by the uplink of a maneuver command if a collision probability threshold is exceeded. This architecture has two fundamental weaknesses in high-congestion environments. First, SSN tracking has significant latency and coverage gaps, particularly for objects below 10 centimeters. Second, communication windows are finite — a vehicle operating autonomously for extended periods cannot depend on timely ground contact to handle a developing threat.
As orbital density continues to increase, the probability of encountering untracked or late-tracked objects grows proportionally. An avoidance system that requires ground uplink for every maneuver decision will eventually face a scenario where latency is fatal. PAVE Space's ADAS was designed from first principles to eliminate this dependency.
ADAS operates as a layered system with three primary components: a passive RF detection array, an onboard conjunction assessment engine, and an integrated maneuver planner coupled directly to the propulsion system.
The passive RF detection array continuously monitors the electromagnetic environment around the vehicle, identifying radar cross-section signatures consistent with tracked debris objects and correlating them against the onboard debris catalog. This catalog is updated via ground uplink during nominal contact windows but can operate in a degraded-but-functional mode for extended autonomous periods using propagated orbital elements.
The conjunction assessment engine runs a continuous short-arc propagation model for all objects within a defined exclusion zone, evaluating relative velocities, miss distances, and collision probabilities on a rolling 72-hour horizon. When a conjunction event is flagged above a defined threshold, the system automatically generates a set of candidate maneuver options and scores them against mission objectives — minimizing delta-v expenditure while maintaining safe separation margins.
The selected maneuver is then executed through the propulsion system with no ground intervention required. Post-maneuver assessment is performed autonomously, and the event is logged for downlink during the next contact window.
ADAS has been validated through an extensive hardware-in-the-loop (HIL) testing campaign conducted over six months at PAVE Space's propulsion and GNC test facility. The campaign included over 1,200 simulated conjunction scenarios spanning a range of orbital altitudes, relative geometries, and debris catalog states. In all tested scenarios, ADAS correctly identified conjunction events above the defined threshold and executed compliant avoidance maneuvers within the required response time budget.
A subset of scenarios intentionally degraded the onboard catalog to simulate extended communication blackouts. ADAS maintained an acceptable false-negative rate under these conditions, demonstrating the robustness of its propagation models in the absence of recent TLE updates.
PAVE Space is currently in discussions with two commercial space situational awareness (SSA) providers to integrate higher-resolution catalog data into the ADAS update pipeline. We are also exploring the integration of LiDAR-based proximity sensing for terminal-phase avoidance of untracked objects — a capability that becomes increasingly important as orbital logistics vehicles approach populated altitude bands. We will share further technical updates as these capabilities mature toward flight qualification.