02.+Code+AF+NAMS-2+Industry+Day+Slides.pdf

Aviation Systems Division Code AF

William Chan

Division Chief

Aviation System Division

NASA Ames Research Center

Major Research and Development Areas

The Aviation Systems Division conducts airspace research to enable the transformation of the National Airspace System (NAS). Research activities include developing concept of operations, fast-time simulations, human-in-the-loop real-time simulations, ATM decision software support tools, flight demonstration and field demonstration deployment support.

Dallas/Fort Worth International 647,834 flights in 2022 3rd among US airports (per ASPM)

American Airlines More than 165 million passengers in 2022 1st among US air carriers

Dallas Love Field 173,856 flights in 2022 33rd among US airports (per ASPM)

Southwest Airlines More than 126 million passengers in 2022 3rd among US air carriers

NTX Research Station…

• NASA research assets embedded in a high-demand, varied operational air transport environment

• Access to ARTCC, TRACON, Towers, 3 air carrier AOCs (American, Envoy and Southwest), and 2 major airports all within 12 miles.

• Supports NASA air traffic research activities from concept development through operational field evaluation.

Fort Worth Air Route Traffic Control Center (ARTCC) (ZFW)

FAA Southwest Region Offices

DFW Terminal RADAR Approach CONtrol (TRACON) (D10)

NASA Ames Research Center North Texas Research Station

Projects and Capabilities

GOAL: Develop, demonstrate, and transition to operations, emerging aviation technologies to identify, monitor, and suppress wildland fires, as a means to enhance safety, improve efficiency, and prevent economic loss.

Advanced Capabilities for Emergency Response Operations (ACERO)

LOGISTICS

AIRCRAFT

SAFETY

SATELLITE

AIR TACTICAL GROUP

SUPERVISOR

AERIAL

IGNITION

DECISION SUPPORT

DAYTIME OPERATIONS NIGHTTIME OPERATIONS

REMOTELY PILOTED

HELICOPTER

REMOTE

SENSING

TANKER

HALE

SEARCH AND RESCUE

WILDFIRE AIRSPACE MANAGEMENT

PILOTED

HELICOPTER

TERRESTRIAL

COMMUNICATIONS

“Incoming Aircraft”

Digital Information Platform (DIP)

Reference Platform and Services

Development of a cloud-based, digital platform and building block services for flight operators and service consumers

University ChallengesNASA Led DIP-Enabled Services for Sustainability

Ground and flight deck services for improving sustainability of aviation operations

Development of innovative AI/ML solution for aviation services

Industry Led Partner Service Evaluations

Integration and demonstration of

Partner services with DIP for validation of the platform “SFNP-Ops Demos” “PS Evals”

SFNP-Ops = Sustainable Flight National Partnerships - Operations

Safe, sustainable, affordable, and accessible aviation for transformational local and intraregional missions

Distribution Center/Warehouse

Inter-City eV/STOL

Inter-city eCTOL

Cargo Delivery

Air Ambulance

Cross-metro Transfer

Airport Transfer

Cargo Delivery

Fleet Operations

Medical Supply Delivery

Small Package Delivery

High Density Corridor

Regional Network

Rural Operations Urban Operations

On Demand Air-Taxi

Advanced Air Mobility (AAM) Subproject Wildfire Fighting

Operations

Pathfinding for Airspace with Autonomous Vehicles (PAAV) Subproject

Crows Landing

X

Simulated* conflict avoidance

KCVH

KWVI

KMRY

KSNS

Hollister

Crewed Segments Autonomous Segments with Crew Supervision

Simulated* contingency, forcing diversion to Hollister

Simulated

* weather avoidance

Visual approach to Visalia via traffic pattern to runway 30

GCS

Notional Demo ScenarioGoal: Enable seamless, scalable, and robust integration of increasingly autonomous UAS cargo operations into the National Airspace System

(NAS)

• Need: Industry has indicated significant interest in regional UAS cargo operations using (e.g. Caravan- or ATR-size aircraft) and a need for remotely piloted (1:1) operations and a pathway to remotely supervised (m:N) operations

• Primary use case is regional, point-to-point cargo transportation served by turboprop and regional jet aircraft with a payload of 1 to 10 tons

Urban Air Mobility (UAM) Subproject

• Motivation o There is no current airspace system to manage scalable UAM operations. The current ATM system relies on traditional NAS protocols with voice-focused ATC which is not scalable

• UAM Objectives o Help ensure the airspace is ready when the vehicles are ready o Empower US-based airspace partners to participate early and take a leading role in spreading the NASA-fostered concepts

• UAM Approach o Investigate the impact of tactical separation provision layer and collision avoidance layer on the safety of UAM operations with various separation standards o Research on interactions between different conflict management layers o Identify vertiport functionality required to support future UAM operations o Advance the demand capacity balancing concept developed for future UAM operations

Evolving the airspace towards integration with the National Airspace (NAS), enabling UAM operations

Extensible Traffic Management (xTM) Subproject

Increase xTM coordination with ATS to improve the flexibility and efficiency of xTM operations xTM and ATS Integration Mature UTM BVLOS Expand to ETM

(Cooperative Operating Practices)

Explore Space Launch & Recovery The Cooperative Operating Practice (COPs) for high-altitude operations based on the UTM construct.

• Operational plan sharing

• Operational plan updating and wait-n-see strategy

• Strategic deconfliction: fallback, agreement, negotiation xTM Maturation and Expansion

AF Software Facility

Freddie - The Federated Airspace Management Framework (Freddie) is a NASA software implementation of an advanced, collaborative, digital, and automated approach to managing air traffic within the airspace.

Extends NASA’s UTM Core capabilities and harmonizes algorithms and concepts within new UTM-inspired aviation domains such as Urban Air Mobility, High-E Traffic Management, Space Traffic Management.

Air Traffic Management Test Bed - Highly scalable, inter-connected simulation and modeling platform to evaluate current and emerging vehicles and airspace management technologies and accelerate the introduction of these advanced technologies in the National Airspace System (NAS).

Sherlock Data Warehouse – enterprise platform for reliable NAS-wide ATM data collection, processing, archival, and delivery leveraging structured Big Data technologies and a tool suite for machine learning and data science. The data includes years of flight, weather, traffic management and facility report data.

WorldWind - NASA-developed, publicly-available, highly-scalable open-source software development kit (SDK) for assisting developers in the creation of geographical information systems (GISs). Its SDK is a powerful tool used by software engineers to build their own GIS applications. WorldWind is used to integrate, analyze, and display a wide-range of airspace, surveillance, weather and other ATM data.

Scalable AI & Sensible ML for ATM

ML based trajectory prediction: Superior performance of ML models that combine physics-based modeling with data-driven methods over traditional physics only models for trajectory prediction.

ML models learn the mapping from prior (historical) flights to their “correct” a/c APM parameters (e.g., drag coefficients) and provide superior ETA predictions.

Retain the physics-based models (for explainability and bounded output behavior) but optimize the model coefficients- thereby exploiting the power of both methods.

Runway Configuration Assistance: Using historical data on runway configuration decisions and prevailing conditions (e.g., surface wind direction), build a ML model to accurately predict the runway configuration (as historically set by the controller/FLMs) and recommend better options (using a RL method).

ML model learns which actions (configurations) are good/bad purely based on historical decisions and outcomes (in terms of delays and traffic flow).

The method is “model-free” and does not require an accurate simulator of surface operations.

ATCSCC Webinar transcription: Accurate automated transcription of FAA command center webinar audio and utilize state of the art NLP/NLU methods to extract relevant context and information from the transcript (e.g., weather related events, delay programs).

With the recent rapid advances made in speech 2 text (e.g., Google), piggy-back on the baseline models and build custom models trained on Aviation language and ATM phraseology.

• Tech pull from FAA NextGen

• Relevant data driven use-cases (e.g., Traffic

Management Initiatives)

• Exploit abundance of Aviation data (e.g., Sherlock), massive computing power (GPUs) and transformer-based ML models (e.g., chatGPT)

• Powered by AF core technology + emerging AI/ML tech + NASA supercomputing resources

Scalable AI & Sensible ML for ATM

ML based trajectory prediction: Superior performance of ML models that combine physics-based modeling with data-driven methods over traditional physics only models for trajectory prediction.

ML models learn the mapping from prior (historical) flights to their “correct” a/c APM parameters (e.g., drag coefficients) and provide superior ETA predictions.

Retain the physics-based models (for explainability and bounded output behavior) but optimize the model coefficients- thereby exploiting the power of both methods.

Runway Configuration Assistance: Using historical data on runway configuration decisions and prevailing conditions (e.g., surface wind direction), build a ML model to accurately predict the runway configuration (as historically set by the controller/FLMs) and recommend better options (using a RL method).

ML model learns which actions (configurations) are good/bad purely based on historical decisions and outcomes (in terms of delays and traffic flow).

The method is “model-free” and does not require an accurate simulator of surface operations.

ATCSCC Webinar transcription: Accurate automated transcription of FAA command center webinar audio and utilize state of the art NLP/NLU methods to extract relevant context and information from the transcript (e.g., weather related events, delay programs).

With the recent rapid advances made in speech 2 text (e.g., Google), piggy-back on the baseline models and build custom models trained on Aviation language and ATM phraseology.

• Tech pull from FAA NextGen

• Relevant data driven use-cases (e.g., Traffic

Management Initiatives)

• Exploit abundance of Aviation data (e.g., Sherlock), massive computing power (GPUs) and transformer-based ML models (e.g., chatGPT)

• Powered by AF core technology + emerging AI/ML tech + NASA supercomputing resources