Small Business Innovation Research/Small Business Tech Transfer
Causal Models for Safety Assurance Technologies
Project Description
Fulfillment of NASA's System-Wide Safety and Assurance Technology (SSAT) project at NASA requires leveraging vast amounts of data into actionable knowledge. Models of accident causation describe a causation chain. The chain would be better understood by examining the large amounts of "everyday" flight data, not just data proximal to high-profile incidents. This proposal is focused on the detection and prediction of more common flight errors or conditions which are necessary for aviation incidents. However, data sets containing safety information are (1) large, (2) distributed, and (3) heterogeneous, making analysis difficult. In order to address these challenges, we propose Causal Models for Safety and Assurance Technologies (CM-SAT). CM-SAT will mine large, distributed, heterogeneous data systems for causal relationships about flight safety. The system will identify causal schema within the data that characterize conditions related to the aircraft and environment that are predictive of failures. CM-SAT will detect causal relationships at varying levels of granularity (e.g. relationships which are unique to a particular flight, to a particular aircraft model, or to a particular fleet). It will leverage state-of-the-art distributed meta-reasoning, which will direct the causal schema learning algorithms to detect and validate causal relationships in different parts of the distributed data sets.
More »
Anticipated Benefits
CM-SAT addresses a relevant, high priority issue – Aviation Safety. The causal relationships learned in this project are directly applicable to the mitigation of aircraft aging, analysis and prediction of crew performance, anomaly detection, and the Verification and Validation of flight control systems. CM-SAT addresses several technical challenge cited by the NASA Aviation Safety Program (AvSP), including the challenges of Assurance of Flight Critical Systems, Discovery of Safety Issues, Vehicle Health Assurance, Crew-System Interactions and Decisions, Loss of Control Prevention, Mitigation, and Recovery, Engine and Airframe Icing, and Atmospheric Sensing & Mitigation. CM-SAT is most relevant to the System–wide Safety and Assurance Technologies (SSAT) project in that it directly addresses safety assurance by predicting risky conditions. However it also indirectly addresses issues in the Vehicle Safety Systems Technologies (VSST) project and Atmospheric Environment Safety Technologies (AEST) by providing general knowledge of the causal relationships on aircraft which lead to hazardous conditions.
CM-SAT analyzes large data repositories for causal relationships that can increase aviation safety, and thus it will be targeted for use by public agencies and private companies interested in aviation safety. The FAA Safety Management System (SMS) would also benefit from actionable safety information, a need which CM-SAT addresses. CM-SAT may be used by military agencies as well, including the Air Force. More broadly, CM-SAT is usable by any agency with large amounts of data that lends itself to causal analysis. Agencies interested in UAS Verification and Validation (V&V), such as ONR, are interested in analyzing collected data to predict performance. The distributed, scalable, heterogeneous causal learning technology that CM-SAT presents is also applicable to the domain of intelligence analysis. More »
CM-SAT analyzes large data repositories for causal relationships that can increase aviation safety, and thus it will be targeted for use by public agencies and private companies interested in aviation safety. The FAA Safety Management System (SMS) would also benefit from actionable safety information, a need which CM-SAT addresses. CM-SAT may be used by military agencies as well, including the Air Force. More broadly, CM-SAT is usable by any agency with large amounts of data that lends itself to causal analysis. Agencies interested in UAS Verification and Validation (V&V), such as ONR, are interested in analyzing collected data to predict performance. The distributed, scalable, heterogeneous causal learning technology that CM-SAT presents is also applicable to the domain of intelligence analysis. More »
Project Library
Primary U.S. Work Locations and Key Partners
| Organizations Performing Work | Role | Type | Location |
|---|---|---|---|
| Aptima, Inc. | Lead Organization | Industry | Woburn, Massachusetts |
Ames Research Center
(ARC)
|
Supporting Organization | NASA Center | Moffett Field, California |
Primary U.S. Work Locations
-
California
-
Massachusetts
Suggest an Edit
Recommend changes and additions to this project record.
This is a historic project that was completed before the creation of TechPort on October 1, 2012. Available data has been included. This record may contain less data than currently active projects.