Session
Advanced Technologies 1
Abstract
Operations engineers face the challenging task of capturing high-value science data while maintaining spacecraft health and functionality and responding to a dynamic space environment using limited information. This task is becoming increasingly complex as missions involve increasingly large numbers of spacecraft, target more elusive science signatures, and operate under growing budget and staffing constraints. Conventional pre-programmed time-sequence commands become increasingly burdensome for multi-spacecraft missions and are fundamentally incapable of reacting to rare, transient science events in real time.
To address this challenge, we have developed MEDOS: the Module for Event-Driven Operations of Spacecraft. Unlike machine learning and neural-network approaches, MEDOS is fully transparent, requires no training data, and is lightweight enough to run on heritage flight processors. Despite this simplicity, MEDOS provides capabilities well beyond traditional rule-based “if-then” autonomy. This paper first describes the mathematical construct underlying MEDOS, which embeds subject-matter expertise directly into onboard decision-making. Physically meaningful derived telemetry parameters are generated onboard and evaluated against fuzzy sets defined by experts to represent events of interest. MEDOS then produces a continuous confidence score indicating the likelihood that a given event is occurring, enabling nuanced and explainable operational responses.
An on-orbit demonstration of MEDOS was successfully conducted on NASA’s flagship Magnetospheric Multiscale (MMS) mission from March 13 to March 20, 2025, raising the system to Technology Readiness Level 7 (TRL-7). A follow-on demonstration is currently being implemented and is expected to run from August 2026 to early 2027. In these demonstrations, MEDOS fuses data from multiple onboard instruments to accurately detect important magnetospheric regions—including the radiation belts, magnetopause, and bow shock—in situ and in real time. This capability enables spacecraft to autonomously respond to a dynamic space environment by adjusting operational modes, prioritizing data for downlink, and safing instruments, when necessary, all without operators in the loop.
Finally, we discuss the application of MEDOS to enable reactive and dynamic spacecraft operations on several upcoming small satellite missions in partnership with the Cal Poly Pomona Bronco Space Lab. On these missions, MEDOS will support real-time detection of the South Atlantic Anomaly and auroral regions to reduce operational burden while preserving mission safety and science return. The results of this work are important for future small satellite missions which must reduce operational burden without sacrificing mission trust or capability.
Document Type
Event
Event-Driven Spacecraft Operations Demonstrated on MMS
Operations engineers face the challenging task of capturing high-value science data while maintaining spacecraft health and functionality and responding to a dynamic space environment using limited information. This task is becoming increasingly complex as missions involve increasingly large numbers of spacecraft, target more elusive science signatures, and operate under growing budget and staffing constraints. Conventional pre-programmed time-sequence commands become increasingly burdensome for multi-spacecraft missions and are fundamentally incapable of reacting to rare, transient science events in real time.
To address this challenge, we have developed MEDOS: the Module for Event-Driven Operations of Spacecraft. Unlike machine learning and neural-network approaches, MEDOS is fully transparent, requires no training data, and is lightweight enough to run on heritage flight processors. Despite this simplicity, MEDOS provides capabilities well beyond traditional rule-based “if-then” autonomy. This paper first describes the mathematical construct underlying MEDOS, which embeds subject-matter expertise directly into onboard decision-making. Physically meaningful derived telemetry parameters are generated onboard and evaluated against fuzzy sets defined by experts to represent events of interest. MEDOS then produces a continuous confidence score indicating the likelihood that a given event is occurring, enabling nuanced and explainable operational responses.
An on-orbit demonstration of MEDOS was successfully conducted on NASA’s flagship Magnetospheric Multiscale (MMS) mission from March 13 to March 20, 2025, raising the system to Technology Readiness Level 7 (TRL-7). A follow-on demonstration is currently being implemented and is expected to run from August 2026 to early 2027. In these demonstrations, MEDOS fuses data from multiple onboard instruments to accurately detect important magnetospheric regions—including the radiation belts, magnetopause, and bow shock—in situ and in real time. This capability enables spacecraft to autonomously respond to a dynamic space environment by adjusting operational modes, prioritizing data for downlink, and safing instruments, when necessary, all without operators in the loop.
Finally, we discuss the application of MEDOS to enable reactive and dynamic spacecraft operations on several upcoming small satellite missions in partnership with the Cal Poly Pomona Bronco Space Lab. On these missions, MEDOS will support real-time detection of the South Atlantic Anomaly and auroral regions to reduce operational burden while preserving mission safety and science return. The results of this work are important for future small satellite missions which must reduce operational burden without sacrificing mission trust or capability.
