You are here

Gaining Actionable Insights Through Neurocomputational Trajectory Segmentation and Clustering

Award Information
Agency: National Aeronautics and Space Administration
Branch: N/A
Contract: 80NSSC21C0551
Agency Tracking Number: 205310
Amount: $750,000.00
Phase: Phase II
Program: SBIR
Solicitation Topic Code: A3
Solicitation Number: SBIR_20_P2
Solicitation Year: 2020
Award Year: 2021
Award Start Date (Proposal Award Date): 2021-07-20
Award End Date (Contract End Date): 2023-07-19
Small Business Information
15400 Calhoun Drive, Suite 190
Rockville, MD 20855-2814
United States
DUNS: 161911532
HUBZone Owned: No
Woman Owned: No
Socially and Economically Disadvantaged: No
Principal Investigator
 (240) 406-5506
Business Contact
 Robin Beahm
Phone: (301) 294-5220
Research Institution

Our NETS solution will apply state-of-the-art neurocomputational methods to partition a trajectory into meaningful segments and then group similar segments into clusters, thus enabling the automatic discovery of common, anomalous, or emergent movement patterns. The initial NETS unsupervised neurocomputational algorithm is able to segment and identify meaningful movement patterns from an aircraft trajectory. The Neurocomputational Model block will be extended to include other data sources such as weather to further develop and refine NETS clustering and anomaly detection capability. In our proposed Phase II effort, we will continue this development but also extend the NETS tool for predicting anomalous movement patterns and enable aircraft trajectory prediction that could include ETA, holding patterns, path stretching, hovering, extreme maneuvering, and non-conformance to nominal patterns. There are two main paths through the NETS architecture: batch processing and real-time streaming analytics. Our proposed NETS tool follows the lambda architecture where there is a batch layer, speed layer and serving layer. The Phase II effort will focus on the batch and speed layer. The batch layer is an extension of our NETS Phase I work where we developed an Aircraft Trajectory Index (ATI) inspired by prior work in Prognostics and Health Management.nbsp; The ATI is a neuro-representation of an aircraft maneuver computed at each segment of the aircraft trajectory. The length of the segment was heuristically determined and set as a fixed length sequence of latitude, longitude, and altitude. We will develop a multi-resolution approach for computing the ATI allowing for varying granularity providing segment level anomaly prediction and extreme maneuvering detection up to detection and prediction of holding patterns, path stretching, and hovering. That batch layer is focused on history and non-real time clustering, detections, and predictions.

* Information listed above is at the time of submission. *

US Flag An Official Website of the United States Government