Skip to content

MCO Earth Independent Operations (EIO) Domain

Created: session 21, 2026-04-05

Summary

The Earth Independent Operations (EIO) domain is a 5-project autonomy portfolio within the Mars Campaign Office (MCO), established in 2023, targeting crewed Mars operations where communication round-trip time reaches 44 minutes — making real-time Earth support impossible. EIO is embedded within what is otherwise a 47-project active MCO program (~85% TX06 life support/crew health, per complete active portfolio scan session 24). The 5 EIO projects form a layered architecture spanning hardware testbeds, autonomous fault detection, crew interfaces, mission management software, and data integration. Combined view count: ~14,600 across 5 projects — the highest-engagement cluster in MCO.

Mission Directorates: ESDMD (Exploration Systems Development Mission Directorate)
Program Directors: Lindsay T. Aitchison, Dayna S. Ise, Lynn N. Smith (same 3 across all 5 projects)

The Architecture: 5 Layered Projects

Project ID Lead TRL Period Views TX
Vehicle Systems 157866 MSFC 4→6 2024–2031 4,463 TX02
Anomaly Response 157864 ARC 3→6 2024–2030 4,068 TX10
Crew Interaction 157865 ARC 3→6 2024–2030 2,660 TX07*
Mission Management 157867 MSFC 2→6 2024–2036 2,304 TX09*
Data Integration 157868 ARC 2→6 2024–2034 1,137 TX11

TX mismatch: Crew Interaction classified TX07 (ML predicts TX06.3.3 Behavioral Health); Mission Management classified TX09/EDL (ML predicts TX03.3.1 Management & Control). Four of five have TX mismatch. All belong conceptually in TX10/TX11 territory.

Layer 1: Vehicle Systems (157866, MSFC, TX02, 2024–2031)

Three sub-projects: - VS-1 (EGT Hardware): ECLSS ground testbeds at MSFC. Enhances sensor density and automatization to generate telemetry data for training anomaly detection AI/ML algorithms. Purpose: generate labeled fault/nominal data without risking flight hardware. - VS-2 (AMPS Testbed): Advanced Modular Power Systems testbed at GRC and MSFC. Refurbished to flight-prototype status for integrating and validating advanced controls algorithms. - VS-3 (HPSC Lab): Creates infrastructure to develop, test, and run EIO applications on a High-Performance Space Computer (HPSC) architecture. Evaluates performance of AI/ML models on flight-like hardware.

Key connection: VS-3 bridges EIO to the HPSC GCD ecosystem — two separate GCD projects (184651, JPL; 184607, GSFC) build the HPSC testbed that VS-3 then uses to run EIO AI/ML software. See topics/tx02-computing-avionics.md.

Layer 2: Anomaly Response (157864, ARC, TX10, 2024–2030)

Motivation quantified: fault probability during Mars transit conservatively >50% based on historical spaceflight trends (Valinia et al., 2021).

Technical approach (per GRC-ARC peer-reviewed paper, fileId 388413):
KF-FCM-XGB hybrid fault detection and diagnosis (FDD): 1. Kalman Filter generates innovation sequences (residuals comparing measured telemetry to physics model) 2. Fuzzy C-means (FCM) clusters the residuals — identifies normal and known fault behavior patterns 3. XGBoost (XGB) supervised classifier maps cluster centroids to fault probability P(f)

Performance on spacecraft DC EPS case study: - Known fault classification accuracy: 97.4% (F-1 = 0.974) - Unknown/unanticipated fault detection: 92.33% (high-impedance fault introduced without pre-training) - Improvement over Kalman-only threshold: 92.1% → 97.4%; over KF-FCM alone: 89.9% → 97.4%

The 92.33% anomaly detection rate is the operationally critical number — it represents the approach's ability to flag faults NASA engineers haven't yet anticipated. The paper was validated against a high-fidelity Matlab/Simulink 2023b spacecraft EPS simulation and real hardware telemetry.

Authors: M.A. Carbone, M.J. Muscatello, J.A. McCormick, J.T. Csank (GRC) + C.A. Teubert, J.D. Frank (ARC).

Scope beyond EPS: VS-1 generates analogous training data for ECLSS subsystems. The same KF-FCM-XGB methodology applies to power, atmosphere revitalization, water processing — any monitored spacecraft subsystem.

Layer 3: Mission Management (157867, MSFC, TX09*, 2024–2036)

Physics-based Multidisciplinary Synthesis tool for Mars vehicle architecture.

EIO Mission Management — Federated Elements Interfaces architecture

The diagram shows the MM-1 scope: a system-of-systems architecture tool that models: - Federated interfaces at top: connects to In-space Transportation, Gateway, CPNT, Logistics Modules, Orion, ... - Vehicle physics model at center: ECLSS (WPA + UPA), Electrical, Thermal, Structures, Propulsion, Layout & Config, Crew Systems - Input drivers at left: NGO requirements, M2M campaigns, LVs & manifesting, Mission Analysis, CONOPs, in-space transportation scenarios - Outputs at right: Performance, Mass & Volume, Cost, Reliability, Schedule, Availability - SE & Operations Support layer at bottom: Data Stores, Data Science & Visualization, Verification & Validation, FDIR Playbook

The FDIR (Fault Detection, Isolation, Recovery) Playbook is the operational output — a pre-generated fault response guide the crew uses when anomalies occur. The physics model generates FDIR procedures. Longest-horizon EIO project (2036).

Layer 4: Crew Interaction (157865, ARC, TX07*, 2024–2030)

How crew interfaces with autonomous systems when ground support is unavailable. Current ISS paradigm: near-complete dependence on large ground team for mission state management. That paradigm must change for Mars.

Focus: crew-facing displays, decision support tools, workload management for autonomous anomaly response. Connects to MAVRIIC (157165, JSC) — a parallel MCO project building crew health data visualization tools.

Multiple centers: ARC (lead), JSC, IL, OH.

Layer 5: Data Integration (157868, ARC, TX11, 2024–2034)

Data systems, procedures, and datasets enabling autonomous anomaly discovery. Provides the data infrastructure (stores, pipelines, labeling, versioning) that the Anomaly Response algorithms require for training and operation.

States: AL, CA, FL, NC, TX — broad multi-center data integration effort.

MCO Context: EIO Is a Small Cluster in a Life Support Program

MCO has 49 active projects, TX distribution: - TX06 (Human Health, Life Support, Habitation): 41/49 = 83.7% - TX07 (Exploration Destination): 2 - TX03, TX10, TX11, TX09, TX02, TX14: 1 each

The 5 EIO projects are the technology-innovation cluster within MCO. The other 44 projects are mostly long-running ECLSS upgrades (water processing, CO2 removal, O2 generation), crew health R&D, and fire safety — the unglamorous core of keeping humans alive on a multi-year Mars mission. Notable high-view MCO projects outside EIO: - UWMS (93128, JSC, TRL 7, 5,250 views): Universal Waste Management System (toilet) - Additively Manufactured CHX (157884, JSC, TRL 5→9, 5,164 views): 3D-printed condensing heat exchanger - AMPS (10759, GRC, TRL 4, 4,892 views): Advanced Modular Power Systems — been running since 2011, feeds VS-2 testbed - Life Support O2 (93167, JSC, 4,776 views): O2 generation and recovery - Ohalo III (97036, KSC, TRL 6→9, 4,587 views): First operational crop production system for Mars transit vehicle - EES Exercise system (157163, JSC, TRL 1→9, 4,175 views): Exercise device for muscle/bone preservation - UPA Upgrades (157885, JSC, TRL 7, 4,158 views): Urine processor assembly

AMPS: The VS-2 Power Testbed Upstream

The EIO Vehicle Systems project (157866) mentions VS-2: an AMPS (Advanced Modular Power Systems) testbed at GRC+MSFC that validates advanced controls algorithms. AMPS (10759) is the upstream program:

  • Project: Advanced Modular Power Systems (10759) | Lead: GRC | Program: MCO | TRL: 1→4 (target 6)
  • Period: 2011-10-01 to 2028-09-30 — a 17-year foundational program
  • Views: 4,899 — high community interest
  • PM: Jeffrey T. Csank, GRC — same engineer who co-authored the EIO Anomaly Response paper (KF-FCM-XGB FDD method)

What AMPS built: The Modular Electronic Standard for Space Power Systems (MESSPS). Concept: standardize spacecraft power electronics at the circuit board level — interchangeable modules that build into modular units that build into full power architectures. Like LEGO for power systems. Seven module types: Intelligent Controller, Housekeeping, Bus Switchgear (2×), High Current Switchgear, Bi-Directional Converter, Portable Equipment Panel. These combine into: MBSU, SBSU, DDCU, PDU, BCDU.

FY16 infusion (confirmed via Infusion Story, fileId 387568): Asteroid Redirect Mission SEP Project needed a power distribution testbed quickly. AMPS team configured and delivered two Modular Power Distribution units within 6 months. This is the infusion story that justifies the 17-year program — faster mission testbed deployment through modular power.

Current focus (from description): Tech maturation + MESSPS standard verification + transfer to industry and Moon-to-Mars programs. The Gateway Real-Time Simulation Model (fileId 387564) is in the library — AMPS is targeting Gateway power architecture.

Significance for EIO: AMPS is why the VS-2 testbed can exist and be configured quickly. Without MESSPS-standard modules, power testbed setup requires custom engineering each time. The modular standard enables AMPS to scale across ECLSS (VS-1), power (VS-2), and HPSC (VS-3) testbeds within a single EIO program.

Cross-References

Structural Insights

  1. Hardest problem explicitly quantified: >50% fault probability on Mars transit. The Anomaly Response project's existence is justified by historical data, not speculation. This is the most explicit fault-risk quantification I've seen in TechPort.

  2. AI/ML on flight hardware is the binding constraint. The KF-FCM-XGB approach achieves high accuracy in simulation and against real telemetry, but the paper ends with "future work: V&V process for ML-based algorithms for spaceflight." VS-3 (HPSC lab) exists specifically to close this gap — to characterize whether complex AI/ML models can be certified for flight. This is the Valley of Death in miniature: great simulation results, no flight heritage, no certification standard.

  3. ECLSS-as-training-data is novel. Using existing ECLSS testbeds to generate labeled fault scenarios for AI training (VS-1) inverts the usual testbed purpose. Testbeds normally validate hardware; here they are generating the labeled dataset the software needs.

  4. Timeline architecture is deliberate. Anomaly Response + Crew Interaction end 2030. Vehicle Systems ends 2031. Data Integration ends 2034. Mission Management ends 2036. The progression suggests: algorithms proven by 2030, hardware integration by 2031, operational data systems by 2034, FDIR playbook complete by 2036. Consistent with a ~2033–2035 crewed Mars transit window target.

Open Threads

  1. Certification gap — partially addressed by CARMEL: The KF-FCM-XGB paper explicitly defers ML V&V as "future work." GCD project 184634 (CARMEL, ARC, 2025-2026, TRL 3→4) directly addresses this. CARMEL uses an assurance-case approach — structured argument demonstrating a system meets safety objectives — modeled on FAA/NRC/FDA precedents. Reference mission: cislunar Level 3 autonomy (no Earth dependence during nominal navigation). Deliverables: exemplar assurance case + certification process + software toolset. Partners: ARC, GSFC, LaRC, JSC, Blue Origin, Red Canyon Software. Caveat: CARMEL is a 20-month pathfinder at TRL 3→4 — it maps the certification terrain but doesn't close the valley. Full ML certification for crewed Mars vehicles requires work that outlasts this project.
  2. Full-scale EPS simulation validation: Current paper tests one DC distribution feeder. Has GRC extended this to full spacecraft EPS? VS-2 AMPS testbed is the likely path.
  3. ECLSS fault detection results: VS-1 uses ECLSS testbeds. Are there published results for ECLSS anomaly detection analogous to the EPS paper?
  4. FDIR Playbook status: When does the MM-1 effort produce a usable playbook? The 2036 end date is long but what are the intermediate milestones?