Establishing a Risk-Based Separation Standard for Unmanned Aircraft Self Separation Roland E. Weibel, Matthew W.M. Edwards, and Caroline S. Fernandes MIT Lincoln laboratory Surveillance Systems Group Ninth USA/Europe Air Traffic Management Research Seminar 14-17 June 2011 ATM Seminar-1 * This work is sponsored by the United States Air Force and Army under Air Force Contract #FA8721-05-C-0002. Opinions, interpretations, recommendations and conclusions are those of the author and are not necessarily endorsed by the United States Government.
UAS Airspace Access Requirement: Sense & Avoid Capability U.S. Federal Aviation Regulation (FAR) General Flight Rules FAR 91.111:...not operate so close to another aircraft as to create a collision hazard FAR 91.113: Vigilance shall be maintained by each person operating an aircraft so as to see and avoid other aircraft [In certain situations] pilots shall alter course to pass well clear of other air traffic Unmanned Aircraft Systems (UAS) inherently lack an onboard pilot to see & avoid other aircraft Technology performs sense & avoid as a means of compliance with see and avoid general flight rules Requirements for sense & avoid performance currently under development Sense & avoid systems must meet demanding safety performance requirements in performing separation functions Risk targets expected to be on the order of 10-7, 10-9 collisions/hr Methods to evaluate safety performance to the required fidelity require analytical performance objectives ATM Seminar-2
North (ft) Sense and Avoid Elements System Elements COTS ground based 3D radar, FAA Airborne sensor development Sensors Data fusion and Tracking Algorithms Threat detection and Maneuver Decision Support Aids Sense and Avoid System Modeling & Simulation P(NMAC relative position) 10000 0.05 8000 6000 4000 2000 0 0.5 0.2 0.1-5,000-2500 0 2500 5,000 East (ft) Standards 1 Airspace Characterization Pilot deviates from clearance Aircraft receiving ATC services UAS pilot deviates from clearance UAS receiving ATC services UAS loses communications link UAS control system fails Fault Tree & Event Trees to Fault Tree Hazard Causes Assess Risk Event Tree Causes Aircraft are proximate ATC services not available ATC services available but fail AND Pilot loses flight control AND OR OR Aircraft on collision course OR Manned aircraft systems fail OR ATC systems fail OR UAS systems fail AND Hazard Loss of separation (Encounter) ATC saves ATC fails Manned See-and-Avoid saves See-and- Avoid fails SS fails Hazard Mitigations Self Separation saves Safety Case MIT Lincoln Laboratory Support Activities Collision Avoidance saves CA fails Consequences Miss Miss Miss Well Clear Violation Midair Collision Open Architectures Testbeds ATM Seminar-3
Sense and Avoid Concepts Ground-Based Airborne Unmanned Aircraft Operational Volume Threat Aircraft Surveillance Volume Unmanned Aircraft Threat Aircraft UAS Ground Station Operating Base Ground-Based Radar Array UAS Ground Station Air Traffic Services Rapid deployment enabled by leveraging existing ground-based surveillance and tools The same surveillance and support hardware supports a diverse range of platforms Operational volume limited by surveillance coverage Unconstrained operations enabled by surveillance volume fixed to platform Longer timelines associated with developing and certifying airborne components Smaller platforms may not have the size, weight, or power to support airborne hardware Path forward: GBSAA ABSAA Hybrid ATM Seminar-4
Well Clear as a Separation Standard General flight rule requirements dictate that aircraft must remain well clear US FAR Part 91 ICAO Rules of the Air Two generally accepted sense & avoid functions Self separation: performance of maneuvers to remain well clear Collision avoidance: aggressive maneuvers to avoid collision Well clear is a standard for performing the self separation function Performance measure for visual separation from other traffic Not previously quantified due to lack of sensor distance measures Consistent with ICAO definition of separation minima * : The minimum displacements between an aircraft and a hazard which maintain the risk of collision at an acceptable level of safety ATM Seminar-5 *ICAO Global Air Traffic Management Operational Concept, Doc 9854, 2005.
Analytical Definition of Well Clear Derived Using Separation Standard Methodology Objective for UAS: separation standard would provide a measurable threshold for UAS sense & avoid safety Sets a clear measure for failure of a function, and supports: design of self separation algorithms/ decision support fast time safety simulation Scalable for future airspace changes beyond UAS Approach: separation standard modeling of risk Assess separation standard relationship to risk of near midair collision (NMAC) Utilize encounter models to determine relationship of relative state and risk over large number of representative encounters Other aspects are also important, but not considered here E.g.: wake vortex, collision avoidance system alerts, etc. Notional Risk-Based Definition of Well Clear Collision Risk Future trajectories Acceptable Risk Relative State between Aircraft 1-2 Aircraft 2 Aircraft 1 ATM Seminar-6 well clear Relative State
Conditional Probability of Near Midair Collision P(NMAC state) Notional example ownship encounter cylinder non-nmac trajectory NMAC region NMAC trajectory 2 trajectories pass through the state shown 1 trajectory results in an NMAC Therefore 1/2 of trajectories through state counted as an NMAC, and 1/2 not counted P(NMAC state) = 0.5 NMAC Cylinder 500 ft 200 ft Note: Modeling performed in 3 dimensions ATM Seminar-7
Encounter Model Used and Associated Assumptions Uncorrelated encounter model: probabilistic model of aircraft dynamics based on surveillance data Highest fidelity encounter model to date of the NAS Based on approximately one year of data from 134 radars from the US Air Force 84th Radar Evaluation Squadron (RADES) Validated framework for TCAS safety studies Allows for large-scale Monte Carlo simulations Derived from 1200-code (VFR) aircraft Assumptions from Models Aircraft randomly blunder into each other Modeling does not include visual acquisition and avoidance maneuvers of either aircraft Modeled P(NMAC) values will be higher than expected ATM Seminar-8
Results Risk contours of P(NMAC state) Distance as state measure: horizontal & vertical views Time to CPA (tau) as state measure: risk curve, mean values TCAS resolution advisories Likelihood of TCAS RA: horizontal & vertical views Risk contours indicate potential well clear boundary definitions If intruder aircraft crosses boundary, it is no longer well clear TCAS alert contours indicate potential interoperability concerns ATM Seminar-9
North (ft) North (ft) Horizontal Position Contours 8000 P(NMAC Relative Position) (example) 8000 P(NMAC Relative Position) 1 0.05 0.9 6000 6000 0.8 4000 2000 0 0.2 0.5 1 0.1 4000 2000 0 0.7 0.6 0.5 0.4 Assumptions 3D simulation 10 million encounter pairs MIT LL uncorrelated encounter model No avoidance mitigations NMAC = 500 ft radius x 200 ft cylinder 0.3-2000 -2000 0.2 0.1-4000 -4000-2000 0 2000 East (ft) -3000-2000 -1000 0 1000 2000 3000 East (ft) 0 ATM Seminar-10
Altitude (ft) Vertical Position Contours 500 P(NMAC Relative Position) 400 300 200 0.01 100 0-100 1 0.5 0.1 0.05 Assumptions 3D simulation 10 million encounters MIT LL uncorrelated encounter model No avoidance actions taken NMAC = 500 ft radius x 200 ft cylinder -200-300 -400-500 -15,000-10,000-5,000 0 5,000 10,000 15,000 North (ft) ATM Seminar-11
Time to CPA Assumptions 3D simulation 10 million encounters MIT LL uncorrelated encounter model No avoidance actions taken NMAC = 500 ft radius x 200 ft cylinder Notes Tau shown in seconds TCAS sensitivity level varies TCAS Version 7.1 ATM Seminar-12
Mean Unmodified Tau Values: Horizontal Assumptions 3D simulation 10 million encounters MIT LL uncorrelated encounter model No avoidance actions taken NMAC = 500 ft radius x 200 ft cylinder Notes Includes both corrective and preventative RAs TCAS sensitivity level varies TCAS Version 7.1 ATM Seminar-13
North (ft) Probability of RA in Effect: Horizontal P(RA in Effect Relative Position) 14000 0.01 12000 10000 0.1 0.05 8000 6000 4000 0.5 Assumptions 3D simulation 10 million encounters MIT LL uncorrelated encounter model No avoidance actions taken NMAC = 500 ft radius x 200 ft cylinder 2000 0 1 Notes Includes both corrective and preventative RAs TCAS sensitivity level varies TCAS Version 7.1-2000 -4000-6000 -10,000-5,000 0 5,000 10,000 East (ft) ATM Seminar-14
Probability of RA In Effect: Vertical Assumptions 3D simulation 10 million encounters MIT LL uncorrelated encounter model No avoidance actions taken NMAC = 500 ft radius x 200 ft cylinder Notes Includes both corrective and preventative RAs TCAS sensitivity level varies TCAS Version 7.1 ATM Seminar-15
Conclusions Analysis of well clear as a separation standard is a viable approach to deriving an analytical definition Builds on ICAO 9689 airspace planning guidance Initial results are promising and straightforward A preliminary definition based on a 5% P(NMAC state) would be: Horizontal: Ellipse 8000 ft ahead, 5,000 ft behind, 3,000 ft laterally Vertical: +/- 300 ft in altitude Continued development of an analytical standard for UAS SAA compliance with well clear is ongoing ATM Seminar-16
Questions Caroline Fernandes Associate Technical Staff Surveillance Systems Email: caroline.sieger@ll.mit.edu Phone: (781) 981-8161 Matt Edwards Associate Technical Staff Surveillance Systems Email: matthew.edwards@ll.mit.edu Phone: (781) 981-4709 Roland Weibel Technical Staff Surveillance Systems Email: roland.weibel@ll.mit.edu Phone: (781) 981-6913 ATM Seminar-17