Andrew Weinert

8 December 2016

Small UAS Well Clear

EXCOM/SARP Multi-agency collaborative sponsorship – FAAPOC: Sabrina Saunders-Hodge, UAS Integration Office Research Division (AUS-300)

DISTRIBUTION STATEMENT A. Approved for public release: distribution unlimited.

Lincoln Laboratory Air Traffic Control Workshop 2016sUAS Well Clear - 2AJW 8 December 2016

Legal Notices

This material is based upon work supported by the Federal Aviation Administration under Air Force Contract No. FA8721-05-C-0002 and/or FA8702-15-D-0001. Any opinions, findings, conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the Federal Aviation Administration.

© 2016 Massachusetts Institute of Technology.

Delivered to the U.S. Government with Unlimited Rights, as defined in DFARS Part 252.227-7013 or 7014 (Feb 2014). Notwithstanding any copyright notice, U.S. Government rights in this work are defined by DFARS 252.227-7013 or DFARS 252.227-7014 as detailed above. Use of this work other than as specifically authorized by the U.S. Government may violate any copyrights that exist in this work.

Lincoln Laboratory Air Traffic Control Workshop 2016sUAS Well Clear - 3AJW 8 December 2016

• Unmanned Aircraft Systems are required to maintain well clear and avoid collisions, in particular with manned aircraft

• Quantitative well clear definition needed for the design and testing of separation systems for small UAS beyond line of sight operations

• This briefing outlines research towards a definition of well clear for small UAS for mid-term concepts of operations at low altitudes– Beyond line of sight in general use airspace– sUAS vs manned aircraft encounters– Out of scope: sUAS vs sUAS encounters and airspaces such as airport

terminals, over heliports or sporting events

Need for Quantitative“Well Clear” Definition for UAS

FAR 91.111: ...not operate so close to another aircraft as to create a collision hazard

FAR 91.113: Vigilance shall be maintained … so as to see and avoid other aircraft … pilots shall alter course to pass well clear of other air traffic

Lincoln Laboratory Air Traffic Control Workshop 2016sUAS Well Clear - 4AJW 8 December 2016

• Detect and Avoid (DAA) Science and Research Panel (SARP) created in 2011to coordinate DAA research– Supports multi-agency UAS Executive

Committee (ExCom)

• SARP well clear working group formedto rapidly deliver recommendation toRTCA SC-228– Lack of well clear definition identified as

highest priority research gap– R&D: August 2013 – August 2014– Multi-agency collaboration: FAA, DOD,

NASA, MITRE, MIT LL,subject matter experts

Previous Effort for Large UASQuantifying Safe Operations

+450 ft

4000 ft

Unmitigated risk threshold: P(NMAC|WCV) = 5%

Tau (time to 4000 ft) = 35 sec

Large UAS well clear adopted by FAA

NMAC = Near Mid-Air CollisionWCV = Well Clear Violation

Lincoln Laboratory Air Traffic Control Workshop 2016sUAS Well Clear - 5AJW 8 December 2016

• Separation standard, based oncollision risk, informed by operational acceptability

• Defined as the relative state where a desired risk threshold is achieved– Unmitigated: regardless of ownship or intruder

avoidance maneuvering– Collision avoidance action likely not needed

• Approach: define risk threshold and map to other states (range, altitude, time, …)

Well Clear as a Separation Standardfor UAS*

Relative State

Collision Risk

Acceptable Risk Threshold

well clear

Aircraft 1

Aircraft 2Relative state between aircraft

Future trajectories

Established Airspace Separation Standard MethodologyApplied to Detect and Avoid

* Applies to unmanned aircraft only

Lincoln Laboratory Air Traffic Control Workshop 2016sUAS Well Clear - 6AJW 8 December 2016

Large UAS Well Clear Definition Process Overview

Define formsand risk

thresholdDefine encounter

modelsTune candidate

definitionsEvaluate

operational suitability SME forum to

down selectcandidates

Make recommendation

SARP review and approval at each step

Primary Roles:

• MITRE, NM State U – Process management

• MIT LL – Monte Carlo simulations and tuning

• NASA – Human in the loop (HITL) simulations

• USAF – Stressing case analysis

• FAA – Subject matter experts (SME)

Lincoln Laboratory Air Traffic Control Workshop 2016sUAS Well Clear - 7AJW 8 December 2016

• Leverage lessons learned from large UAS well clear research• Approach:

– Define desired unmitigated risk = P(NMAC|Well Clear Violation)

– Explore different definition forms• Existing (Large UAS) form: mod-tau*, horizontal and vertical thresholds• “Hockey Puck” based on horizontal and vertical thresholds

– Tune parameters to risk threshold• Based on M&S using sUAS CONOPS and dynamics in operational airspace

– Consider operational suitability for sUAS operators and airspace users

Defining Well Clear for sUAS

Community Objective: Develop a sUAS vs manned aircraft Well Clear definition based on risk and operational suitability

* Time to horizontal distance thresholdNMAC – Near mid-air collision (HMD ≤ 500 ft, VMD ≤ 100 ft)

Lincoln Laboratory Air Traffic Control Workshop 2016sUAS Well Clear - 8AJW 8 December 2016

• Well Clear Overview

• Modeling and Simulation

– M&S Components

– Representative sUAS

– Manned Aircraft

– Simulation Environment

• Results and Recommendation

Outline

Lincoln Laboratory Air Traffic Control Workshop 2016sUAS Well Clear - 9AJW 8 December 2016

Modeling & Simulation ComponentssUAS vs. Manned Aircraft

Buildings

Airspace Class

Weather

Birds

People

General Aviation(Part 91)

CommercialAviation

(Part 121)

Large UAS

Study Focus

Potential Considerations

sUAS Swarm

sUAS

Helicopter Air Ambulance

Airspace Structure

Addressing challenges associated with limited sUAS flight data and

limited manned flight data at altitudes below 1200 ft. AGL

Terrain & Vegetation

Lincoln Laboratory Air Traffic Control Workshop 2016sUAS Well Clear - 10AJW 8 December 2016

Risk Assessment Architecture:Monte Carlo Fast Time Simulations

Development Focus

Rawradar data

Tracking and fusion Featureextraction

Collision Avoidance and Self-separation

Algorithms

Collisions per encounter

Encountermodels

Relative Risk analysis

Fast-timesimulation

Surveillance models

Aircraftflight profilesand dynamics

Collision Avoidance Safety System Tool (CASSAT)

LL Grid Computing EnvironmentCompute Nodes 274

Compute Cores 8768Peak Performance

77.1(TFLOPs)RAM (TB) 17.5

Central Storage (TB) 1,200

Distributed Storage (TB) 2,466

Network Storage

Compute Nodes Service Nodes Cluster Switch

LAN Switch

Shared File System

Scheduler

Monitoring System

Login Node

Previous Assessments

• Developed to certify TCAS version 7.1 • SARP large UAS well clear definition• Evaluation and tuning of multiple UAS

algorithms

Lincoln Laboratory Air Traffic Control Workshop 2016sUAS Well Clear - 11AJW 8 December 2016

ID 1 2 3 4MGTOW (lb)* 0 - 4.4 0 - 20 0 - 20 20 - 55

Mean Cruise Airspeed (kt) 25 20 30 60

Max Airspeed (kt) 40 30 60 100

Descend Rate (fpm) -300 -500 -500 -1000

Climb Rate (fpm) 500 700 700 1000

Representative sUAS55 lbs. while flying below 1200 ft AGL

ScanEagleDJI Phantom 4

Trajectories

Vertical Transit Horizontal Transit SectorSpiralCreeping Line

Aeromapper 300GoPro Karma

Notional sUASExamples

AGL – Above Ground LevelMGTOW– Max Gross Take-Off Weight

Lincoln Laboratory Air Traffic Control Workshop 2016sUAS Well Clear - 12AJW 8 December 2016

• Wide range of manned aircraft behavior simulated using two different probabilistic Bayesian network encounter models

• Uncorrelated fixed-wing encounter model – Represents low-altitude general aviation behavior– Since 2008, has supported a wide-range of manned

and UAS safety studies

• Helicopter air ambulance encounter model – Represents at-risk low altitude helicopter

operations– Newly developed from anonymized FOQA records

of Boston-area Medevac flights from 2015 – 2016– First helicopter focused statistical encounter model

Manned Aircraft ModelingFixed-Wing and Helicopters

MIT LL Uncorrelated Fixed Wing Encounter Model

FOQA – Flight Operational Quality Assurance Data

Lincoln Laboratory Air Traffic Control Workshop 2016sUAS Well Clear - 13AJW 8 December 2016

sUAS• 4 Representative Platforms• 5 Trajectory Types

Uncorrelated Manned Aircraft• 1200 code Fixed-Wing• Helicopter Air Ambulance

Airspace• Airspace Class (B,C,D,E,G)• Region (CONUS, Offshore)

Well Clear Forms• Spatial (i.e. Vertical miss distance)• Temporal (i.e. Tau)

Summary of Sensitivity Parameters

6+ million encounters simulated to derive the unmitigated risk: Sensitivity of NMAC risk to various parameters analyzed.

Lincoln Laboratory Air Traffic Control Workshop 2016sUAS Well Clear - 14AJW 8 December 2016

• Well Clear Overview

• Modeling and Simulation

• Results and Recommendation

– Example encounter

– Sensitivity analysis

– Risk contours

Outline

Lincoln Laboratory Air Traffic Control Workshop 2016sUAS Well Clear - 15AJW 8 December 2016

• Encounter representative of DJI Phantom sUAS and Cessna 150

• Manned aircraft flies 5X faster

• sUAS has minimal ability to influence outcome

• Likely well clear violation

Influence of Aircraft Speed Difference Encounter Example

Start CPA Turn

Y (ft

)

X (ft)0

-5000

-4000

-3000

-2000

-1000

0

1000

2000

3000

4000

5000

CPATime = 49 sRange = 930 ft

Manned

sUAS

StartTime = 0 sRange = 5626 ft

sUAS MannedAirspeed 15 kt 81 kt

Distance traveled from start to CPA 1177 ft 6580 ft

CPA – Closest point of approach

Lincoln Laboratory Air Traffic Control Workshop 2016sUAS Well Clear - 16AJW 8 December 2016

• Adding temporal criteria to HMD and VMD filter has limited influence on P(NMAC| Well Clear Violation)– Attributed to sUAS’ relative slow airspeed

• Recommend sUAS well clear definition to only use (HMD, VMD)

Sensitivity Analysis Example:Inclusion of Temporal Variables

Spatial Only

HMD – Horizontal miss distance (ft) VMD – Vertical miss distance (ft)NMAC – Near mid-air collision (HMD ≤ 500 ft, VMD ≤ 100 ft)

Spatial + TemporalVM

D (f

t)

HMD (ft)1000 2000 3000 4000

100

200

300

400

450

VMD

(ft)

HMD (ft)1000 2000 3000 4000

100

200

300

400

450

Example: 10% chance of an NMAC if

HMD = 2000 ft &VMD = 250 ft

Lincoln Laboratory Air Traffic Control Workshop 2016sUAS Well Clear - 17AJW 8 December 2016

• NMAC risk has limited to no sensitivity to the sUAS’ trajectory– sUAS in vertical transit has no horizontal ability to reduce risk– At slow airspeeds, sUAS simply can’t travel far during an encounter

• Unmitigated risk is not sensitive to a sUAS trajectory

Sensitivity Analysis Example:sUAS Trajectories

Vertical Transit Horizontal Transit Pattern

VMD

(ft)

HMD (ft)5001000 2000 3000 4000

100

200

300

400

450

VMD

(ft)

HMD (ft)5001000 2000 3000 4000

100

200

300

400

450

HMD – Horizontal miss distance (ft) VMD – Vertical miss distance (ft)NMAC – Near mid-air collision (HMD ≤ 500 ft, VMD ≤ 100 ft)

VMD

(ft)

HMD (ft)5001000 2000 3000 4000

100

200

300

400

450

Lincoln Laboratory Air Traffic Control Workshop 2016sUAS Well Clear - 18AJW 8 December 2016

• “Hockey puck” well clear definition using HMD and VMD– 250 ft. VMD x 2000 ft. HMD– ~ Half height and distance of

large UAS well clear– No temporal conditions

• Risk consistent across CONOPS, sUAS and manned aircraft dynamics models

• Subject to additional validation

• Full technical paper in progress

sUAS vs Manned AircraftSARP’s Well Clear Recommendation

VMD

(ft)

HMD (ft)

P(NMAC|WCV) (%)

500 1000 1500 2000 2500 3000 3500 4000100

150

200

250

300

350

400

450

Average risk contours over all models

SARP’s Recommendation to FAA ExCom

HMD – Horizontal miss distance (ft) VMD – Vertical miss distance (ft)NMAC – Near mid-air collision (HMD ≤ 500 ft, VMD ≤ 100 ft)

Lincoln Laboratory Air Traffic Control Workshop 2016sUAS Well Clear - 19AJW 8 December 2016

• Well clear definition for sUAS vs manned aircraft needed to extend operations to beyond line of sight– Definition based on unmitigated risk and operational suitability

• Risk modeling required development of new low altitude encounter models– Representative sUAS platforms and CONOPS– New model for low altitude helicopter operations

• Risk determined to not be sensitive to assumptions– Lack of sensitivity attributed to generally slow sUAS airspeeds– Additional analyses recommended to evaluate niche cases such as consistently

fast sUAS operations

• Additional community vetting of operational suitability expected

Summary

SARP’s recommended well clear definition: “hockey puck” centered on sUASHorizontal 2,000 ft, Vertical 250 ft

Lincoln Laboratory Air Traffic Control Workshop 2016sUAS Well Clear - 20AJW 8 December 2016

Thank You

Andrew WeinertAssociate Technical StaffHumanitarian Assistance and Disaster ReliefEmail: andrew.weinert@ll.mit.eduPhone: (781) 981-0986

Questions?

Feedback?

Rodney ColeAssistant Group LeaderSurveillance SystemsEmail: rodc@ll.mit.eduPhone: (781) 981-7423