NASA Headquarters Earth Independent Operations (EIO) Partnerships in Anomaly - Request for Information (RFI)

Amendment #1: The response date for this RFI has been extended to 11:59PM EST, February 18th, 2026.

NASA Headquarters (HQ) is hereby soliciting information from potential sources for Earth Independent Operations Partnerships in Anomaly.

NASA HQ is seeking capability statements from all interested parties, including all socioeconomic categories of Small Businesses and Historically Black Colleges and Universities (HBCU)/Minority Institutions (MI) for the purposes of determining the appropriate level of competition and/or small business subcontracting goals for (insert acquisition title). The Government reserves the right to consider a Small, 8(a), Women-owned (WOSB), Service-Disabled Veteran (SD-VOSB), Economically Disadvantaged Women-owned Small Business (EDWOSB) or HUBZone business set-aside based on responses received.

This Request for Information (RFI) does not constitute a commitment, implied or otherwise, that NASA will take action in this matter.  This RFI is for US domestic entities only.  Interest from international government agencies or other international entities will be addressed in direct conversations with NASA Headquarters.  International inquiries should be addressed to the Point of Contact in Section 6.

No solicitation exists; therefore, do not request a copy of the solicitation. If a solicitation is released, it will be synopsized on SAM.gov. Interested firms are responsible for monitoring this website for the release of any solicitation or synopsis.

Introduction

The NASA Exploration Systems Development Mission Directorate, through its Mars Campaign Office, seeks to revolutionize human spaceflight operations for missions experiencing significant communications delays or outages by developing the technology necessary for human crew to operate fully independently of Earth-based ground support for short periods of time during mission-critical operations.   The research and development of an integrated set of technologies to enable safe and independent crewed operations is the responsibility of the Mars Campaign Office and managed through the MCO Earth Independent Operations portfolio.

MCO is investigating potential commercial partnerships in the development and scaling of EIO technologies.  A series of targeted RFIs will be released to scope potential responses and allow for effective review and subsequent engagement.  This RFI specifically targets technologies in the “Anomaly Response” portfolio of projects, which focuses on the use of novel computing methods including Artificial Intelligence and Machine Learning methods to support both crew-led diagnostics for rapidly evolving hazardous conditions and crew-led development of response plans to mitigate operational anomalies unpredicted during design and testing.

Written responses from domestic organizations including industry, academia, investors, philanthropic organizations, and other stakeholders and partners, as well as other federal agencies, state, and local governments, are encouraged.  Information may be submitted by any domestic organization or individual.  There is no limit to the number of responses an organization may submit to this RFI.  Responses to this RFI will be kept strictly confidential.

Purpose

The objective of this RFI is to invite community responses on the availability and scope of existing technology relevant to the scope of work described in the Subsection 2.1 below and capable of operating subject to the constraints listed in Subsection 2.2. MCO seeks information on technologies that are sufficiently well-established to show relevant results in commercial applications but that may not be specifically targeted on spaceflight use cases.  MCO seeks information on facilities or laboratory capabilities that may be useful for the testing of Anomaly Response technologies.  MCO further seeks information on organizations with a workforce that has a proven track record in delivering technology directly related to the scope of work described.

Scope of Work

Background

Future crewed Mars missions will experience significant communications delays and potential blackouts, limiting the availability of timely ground assistance during short and medium time-to-effect mission-critical anomalies. To ensure crew safety and mission success, the vehicle and crew must be capable of effectively and promptly responding to unanticipated or ambiguous faults independently for limited periods.

Anomalies of interest include, but are not limited to, rapidly evolving hazards in life support, power, thermal, avionics, propulsion, guidance/navigation/control, communications, and vehicle structures/fluids that were not fully predicted in design and test. There is specific interest in technologies and approaches that are applicable across systems and target configurations.

The Anomaly Response portfolio seeks to advance a cohesive set of on-board capabilities that combine physics-based reasoning, knowledge-driven methods, and data-driven AI/ML to achieve the Anomaly Response Objectives (See 2.1.2 Objectives).

The portfolio emphasizes approaches that are robust to limited compute, intermittent data, complex systems, and evolving configurations; provide human-understandable rationale; and integrate with flight software, fault management, and autonomy frameworks.

The Anomaly Response portfolio is currently divided into five projects:

1. System Diagnostics: anomaly detection, feature engineering, known fault isolation and ranking
2. Fault Hypothesis Generation: unknown fault isolation, fault inference, uncertainty estimation, combining results from multiple diagnostic tools
3. Procedure Synthesis: Generating and optimizing novel procedures for fault diagnosis, mitigation, safing, and resolution
4. Procedure Validation: Validating that novel procedures do not endanger crew, mission-success, or unnecessarily reduce redundancy beyond mission-success (e.g., spares)
5. Crew Decision Support: Ensure crew has diagnostic information and situational awareness to inform selection of appropriate fault response

The portfolio is using a combination of synthetic (i.e., simulated), testbed, and ISS data for model and technology training and validation. Technologies are being tested on a combination of power distribution systems, Urine Processing Assembly (UPA), and the 4Bed CO2 scrubber. Note that these were chosen as evaluation systems, but technology is being developed to apply beyond these.

NASA anticipates that robust Anomaly Response capabilities will have significant terrestrial applications (e.g., aviation, energy, manufacturing, automotive, data centers) and encourages responses that consider both spaceflight and dual-use potential.

Objectives

MCO invites responses addressing any subset of the following primary objectives. Respondents should identify the objective(s) addressed and provide evidence of maturity in operational or operationally relevant settings.

Primary objectives:

1. Identify anomalous patterns from streaming telemetry, video, audio, images, or other data sources in real-time
2. Generate and prioritize diagnostic hypotheses at various scales (e.g., Fault Mode A, Unknown Fault in Component B, Unknown Electrical Fault in Subsystem C)
3. Generate inferences about the nature of an unknown fault that can aid in identifying an effective fault response (e.g., similar faults, isolate to systems, correlate with system states, etc.).
4. Predict, with likelihood, potential downstream effects of a fault, including safety thresholds, damage to system, hazard to crew, impact to mission success, or other events of interest.
5. Estimate, with uncertainty, time to effect for each potential effect.
6. With crew input, synthesize a novel response procedure (e.g., to further isolate a fault, safe a faulty system, mitigate effects, etc.) when no pre-certified procedure exists.
7. Validate candidate procedures for safety and effectiveness prior to crew execution.
8. Present diagnostic and procedure information (estimates, uncertainty, telemetry, procedure options, rationale, risks, etc.) to support crew decision-making under time pressure and high cognitive load.

Supporting objectives:

1. Incorporate crew input into diagnostic process (e.g., unstructured observations “I smell smoke”, instinct “I don’t think It’s A”, or structured results “test A is positive”) and procedure synthesis process (mixed-initiative planning)
2. Validate with Mars-relevant human-in-the-loop scenarios
3. Integration with flight software and compatibility with High-Performance Space Computing (HPSC)

Architecture

The Anomaly Response software architecture is a subcomponent of the larger, integrated, EIO software architecture, which runs onboard or local to the vehicle. The Anomaly Response (AR) stack provides the crew with AI-enabled diagnostics and response planning. It is modular, consisting of elements for diagnostics, procedure synthesis, procedure validation, and crew decision support tools. The existing EIO architecture leverages real-time telemetry, mission history (maintenance history, past telemetry, engineering data, crew/ground input to perform these tasks.

This architecture is deliberately composable and subsystem-agnostic. It enables use cases such as electrical load monitoring—treating power as a “heartbeat” that reflects the health of connected systems—while remaining extensible to life support, thermal control, and other domains. By combining multiple methods rather than relying on a single “silver bullet,” AR delivers resilient, explainable support that enhances crew autonomy and mission safety.

MCO welcomes inputs that supplement the existing EIO architecture as well as those that propose an alternative or improvements.

Relevant Technologies

MCO welcomes information on technologies, methods, and facilities that enable the objectives above. Note that responders are not expected to include all these technologies in their response and the list is not exhaustive. Examples include, but are not limited to:

• Anomaly detection
  • Time-series methods, neural networks, clustering techniques, physics residual-based approaches, physics-informed ML, hybrid “grey-box” model-based and ML fusion, ensemble methods and Mixture of Experts, hierarchical systems approach, outlier detection, density-based methods, advanced rule-based approaches, multi-modal methods (e.g., combining images and time-series), fuzzy logic, online learning, methods leveraging historical data, methods leveraging engineering/design data, methods leveraging crew/ground input, sequence models and foundation models for multivariate telemetry and event streams (potentially including multi-modal data such as telemetry, text, images, video, and audio).
• Diagnostics and fault inference
  • Model-based reasoning (e.g., dependency graphs, FMEA/FTA-informed models), Bayesian inference, constraint-based reasoning, deductive reasoning, inductive reasoning, causal discovery/causal graphs, case-based reasoning, pattern recognition, multi-modal methods (e.g., combining images and time-series), hierarchical systems approaches, optimization-based methods, fuzzy logic, online learning, methods leveraging historical data, methods leveraging engineering/design data, methods leveraging crew/ground input, sequence models and foundation models for multivariate telemetry and event streams (potentially including multi-modal data such as telemetry, text, images, video, and audio), graph representation learning and graph neural networks (GNNs) over system topologies, dependency graphs, and knowledge graphs.
• Feature engineering
  • Dimensionality reduction, relational features, preprocessing techniques, multi-modal data fusion, spectral multiscale features, temporal structures, calibration drift compensation, outlier detection, virtual sensors
• Procedure synthesis and planning
  • Automated planning (e.g., HTN, temporal planners), constraint programming, policy search, goal/option discovery, rule learning, LLM-augmented planning with safety guardrails and verification, mixed initiative planning methods, probabilistic methods, safe and constrained reinforcement learning, including offline RL from logged operational data and simulation, with explicit safety constraints and verification
• Procedure validation and assurance
  • Formal methods (model checking, reachability/viability analysis), runtime assurance, conflict detection, hazard analysis automation, Monte Carlo with importance sampling, discrete event simulation, safety case/assurance case generation, ergonomic/human factor assessment, resource planning and utilization assessment, stochastic petri nets, assessment over disassembly bill of materials, advanced control flow analysis, attribute grammars, insight generation, static analysis, semantic state trees, compiler logic and technologies, hybrid Bayesian networks, Markov processes
• Time-to-effect and prognostics
  • Remaining-useful-life estimation, hazard progression models, reduced-order physics models, stochastic hybrid systems, machine-learning methods, physics-informed machine learning, hybrid “grey-box” data-driven/machine learning, uncertainty propagation, resource accounting, causal/dependency graphs, state space observers with forward simulation, dynamic Bayesian networks
• Crew decision support and HSI
  • Mixed-initiative interfaces, explanations and rational generation methods, saliency/attribution methods for ML, adaptive alerting and workload management
• Physics modeling and digital twins
  • Reduced-order and acausal modeling, domain-specific models for ECLSS/power/thermal/fluids, techniques for low-compute inference, degradation-modeling frameworks, generative models (e.g., variational methods, diffusion models, normalizing flows) and simulation-based inference leveraging high-fidelity or differentiable simulators, within compute constraints.
• Expert systems and knowledge engineering
  • Rule engines, knowledge graphs, ontologies, hazard/rule capture and maintenance, provenance management.
• Machine learning under constraints
  • Scheduling and real-time considerations, robust ML against shift/drift and low data, robust ML against shift/drift and low data, including continual learning, transfer learning/domain adaptation, and (where appropriate) federated or distributed learning.
• Verification, validation, and test
  • Machine learning verification and validation methods, performance metrics, operator performance evaluation methods, benchmarking, safety and assurance frameworks specific to data- and ML-driven systems (e.g., ML assurance cases, data/model documentation, robustness and interpretability testing).

Requested Information

This is an RFI for Earth Independent Operations partnerships in the Anomaly Response portfolio.  Evaluations of capability statements will not be issued to respondents.

Response Format

All responses must be received by February 18, 2026. Responses are limited to a maximum 10 pages in length using 12-point Times New Roman font style for the main text, single space pages with one-inch margins.  A 9-point font may be used for text within figures, tables, and charts.  Note that proprietary or export-controlled information may be included, provided it is clearly marked and properly protected according to applicable regulations.

Responses must be submitted in Adobe PDF format.  The total file size for an individual submission is limited to 10MB.

Response Content

Responses should address at a high level the following aspects of the technology or capability presented.

1. Title and Technical Description
   
   The response should provide a clear title and detailed technical description of the capability.  Include sufficient detail to identify key technology elements and describe at a high level any elements that are proprietary or restricted in nature.  The response should not provide detail on the methods by which those elements operate but should identify the functionality provided and the input.
    
2. Identification of Response Type
   
   The response should identify whether it describes a technology, a facility, a workforce element, or a combination of multiple elements using the definitions in Section 2.  Short descriptions or clarifications of the response type are encouraged if appropriate.
    
3. Identification of Relevant Anomaly Response Objective
4. Identification of Technology Elements
5. Current Maturity
6. Existing Commercial Use Cases and Development Timelines
7. Description of Results from Prior Use or Testing
8. Collaboration Opportunities

Information for Respondents

This information is requested for planning purposes only and does not constitute a solicitation.  The release of this RFI does not obligate the government to issue a future solicitation nor does it obligate the government to invest any resources to any specific space technology topic area.

Respondents are encouraged to provide information that is not constrained by limited or restricted data rights.  No Personally Identifiable Information (PII) should be submitted in response to this RFI.  Respondents are solely responsible for all expenses associated with responding to this RFI.  NASA intends to consider all data received to inform the Agency’s future planning. 

Please note that NASA employees and other government agencies as well as its support contractors’ employees and/or their subcontractors working on behalf of NASA may review respondent’s submissions.  NASA contractors and subcontractors are governed by non-disclosure provisions in their applicable contracts and subcontracts, which protects the confidentiality of all information reviewed.  Responses to this RFI will not be returned, and respondents will not be notified of the results of the review.

This RFI is for planning purposes only and shall not be considered as an obligation on the part of NASA to acquire any products or services and/or enter into any partnership agreements or any other legal implementing instrument as a result of submissions received through this RFI.  NASA will not pay for the information requested in this RFI.  The information provided is entirely voluntary and does not affect the ability to propose on future opportunities.
 

Points of Contact:

Andres Martinez
Domain Lead, Earth Independent Operations
NASA Headquarters
Exploration Systems Development Mission Directorate

Submitting Responses

Responses to this RFI must be submitted electronically to Andres.Martinez@nasa.gov by 11:59 Eastern Time on February 18th, 2026.

In your response, please provide the following:

• Name, Address, and Point of Contact for the responding organization
• The response in electronic copy as a single searchable and unlocked document in the Portable Document File (PDF) format following the requirements described in this solicitation.
• Responses including references, studies, research, or other empirical data not widely published shall include copies or electronic links of the referenced materials.  In the event that a link is provided, it is the responsibility of the responder to ensure that NASA reviewers are able to access the desired information.