Project on Unmanned Aircraft in the NAS Final Review Panel Meeting 1 Integration of Unmanned Aircraft into the National Airspace System A Project Course by Carnegie Mellon University Dept. of Engineering and Public Policy Dept. of Social and Decision Sciences May 1, 2007 2 Expert Review Panel Tom Curtin, AUVSI Bret Davis, AUVSI Lexa Garrett, America West Airlines Jim Geibel, GAO David Gerlach, FAA Tom Henricks, Aviation Week Ramon Lopez, Aurora Flight Sciences Edmond Menoche, GAO Rene Rey, FAA Melissa Rudinger, Aircraft Owners & Pilots Assn. James Sizemore, FAA Larry Thomas, GAO Dyke Weatherington, DoD/OSD 3 Purpose of CMU Project Courses in Technology and Policy Analyze a “real world” policy problem involving technology Combine diverse information and analytic frameworks to derive policy insights Learning objectives: Problem decomposition, structuring and formulation Interdisciplinary problem solving Communication Teamwork 4 Examples of past project courses Title Year Safety and the Light Truck Craze: Who Wins? Who Loses? Who Cares? 2000 Environmental Impacts of E-commerce - A case study of book purchasing 2000 Sustaining Pittsburgh's Vital Services When the Power Goes Out 2004 Wireless Communications Systems for Emergency Responders 2004 Hybrids and Diesels in the American Automobile Fleet 2005 The Impact of Spyware 2005 Safety of Implanted Cardiac Devices 2007 5 Contributors to our UAS project 20 undergraduates majoring in: Engineering Social Science Business Administration 3 Ph.D. student managers 3 faculty advisors Expert review panel Other experts 6 Background for this Project Increasing demand for UA Military (many current uses) Civilian (many potential uses) Federal Aviation Administration (FAA) is developing a roadmap for integrating UA into the NAS A few of the issues to be addressed: Safety and reliability Public acceptability Market viability 7 Analysis Areas Economics How cost-effective are UA compared to alternative means of providing specific services? Risk, Technology and Standards What are the regulatory implications of different approaches to “equivalent level of safety?” Public Awareness and Perceptions Are risks of UA of greater public concern than risks of manned aircraft? Governance How can the current system for deliberation and decision-making on UA access be improved? 8 Project Outcomes ~16 person-months of research completed across the four focus areas Economic model of market viability Risk model of fatality implications of UA introduction Better understanding of public awareness & risk perception Actor & “roadblock” analysis yields insight on deliberative process for UA integration Regulatory & policy recommendations 9 Economics Team Members: Nathan Diorio-Toth Feng Deng Reiko Baugham Victoria Morton Brad Brown Team Manager: Ryan Kurlinski 10 Purpose Assess the market viability of UAS applications using relative cost effectiveness Assess the effect of various regulatory measures on the market viability of UAS applications 11 Goals Develop UAS cost model Cost components Airframe Communications Insurance Pilot Etc. Apply cost model to chosen applications and alternatives to compare cost Examine sensitivity of overall cost to changes in each cost component Estimate cost implications of different regulatory measures and technology improvements 12 UAS Applications Weather Reconnaissance Alternative: WC-130J Hercules: high-wing, medium range aircraft Pipeline monitoring Alternative: Concentric sensors: pressure sensitive sensors Localized Surveillance Alternative: Traffic Helicopter: e.g. Bell JetRanger 13 Analysis Method Used triangular distributions to assign probable ranges to each input cost From this, generated a Probability Density Function Probability Density Function shows the entire range of possible costs with the associated likelihood of each cost Allows analysis of the most probable cost advantages 14 Importance Analysis Contribution of uncertainty in each input to uncertainty in total cost Triangular probability distributions of all input variables Economic Model 15 Weather Reconnaissance Analyzed the use of Aersonde UAS for Weather Reconnaissance vs. the use of the WC-130J Hercules Aerosonde UAS currently in use for Weather Reconnaissance Hercules WC-130J currently in use by Keesler Air Force Base 16 Results: Weather Reconnaissance Probability Density 12u Probability Density of UAS Cost Advantage 10u 8u 6u 4u 2u ($/flight hour) 0 0 100K 200K 300K 400K 500K 600K 700K Alt Cost-UAS Cost 17 Importance in Alt Cost-UAS Cost Results: Weather Reconnaissance Importance Analysis of Model Inputs 1 0.8 0.6 Mission Hours per Year Operational Lifetime Com-Link Cost 0.4 0.2 0 Component costs Manpower Com-Link Cost Operational lifetime Cost per gallon Insurance Rate Safety Technology Cost Hours per year Gallons per Hour Alt Cost-UAS Cost Inputs 18 Results: Weather Reconnaissance Key Results: UAS more cost effective than current manned alternative Most important inputs in determining overall cost effectiveness: Mission hours per year Com link cost Operational lifetime Currently available sense-and-avoid equipment cause significant decrease in cost effectiveness, but does not cause the UAS to be more expensive than the manned alternative 19 Pipeline Monitoring Analyzed the use of the Aero Environment AeroPuma vs. the use of concentric wire sensors ($6+/m) Note the difference in monitoring style UAS monitors using thermal imaging with each pass and relays pertinent leak info to docking stations Concentric sensors constantly monitor pipeline and relay information 20 Results: Pipeline Monitoring Key Results: UAS cheaper depending on number in use Important to note difference in monitoring styles between UAS and concentric sensor Important inputs: Relay/Docking station cost Number of UASs in use 21 Localized Surveillance Application based on the surveillance of a 1km2 area for a short time (~1-3 hours) Considered the use of a Cyber Defense Systems CyberBUG vs. the use of a traffic helicopter For model inputs, considered monitoring a large traffic accident over 2 hours For policy considerations, analyzed the addition of mandated sense-and-avoid hardware to the UAS 22 Results: Localized Surveillance Probability Density PDF of Cost per Mission for UAS Compared with Manned Alternative Note: no meaningful overlap 2000 4000 6000 8000 10K Cost per Mission ($) 12K 14K 16K 23 Probability Density Results: Localized Surveillance PDF of Cost per Mission for UAS Compared with Manned Alternative with High-Range Fixed Cost Variance Note: still no meaningful overlap 2500 5000 7500 10K 12.5K 15K 17.5K 20K 22.5K Cost per Mission ($) 24 Results: Localized Surveillance Probability Density PDF of Cost per Mission for a Larger UAS Capable of Carrying Sense-and-Avoid Equipment Compared with the Cost of Manned Alternative Note: Significant overlap indicating that UAS would likely no longer be a viable alternative to manned craft 0 10K 20K 30K 40K 50K 60K 70K Cost per Mission ($) 25 Input Importance for Cost Per Mission Results: Localized Surveillance Missions per Year Mission Related Costs Flight Hours Per Mission Importance of inputs. Input Costs 26 Results: Localized Surveillance Key Results: UAS less expensive in almost every case Levelized cost for manned more sensitive than to utilization hours & discount rate than cost for unmanned UAS cost effectiveness reduced significantly by requirement for sense-and-avoid hardware Important inputs: Missions per year Discount rate Flight hours per mission 27 Policy Implications Analyzed the effect of the following policies: Mandated insurance premiums Mandated use of A/N hardware (Increased fixed cost) Mandated record-keeping practices (Increased yearly cost) Mandated airframe materials (Increased fixed cost) Mandated minimum amount of safety equipment (Increased fixed cost) Mandated pilot/operator training 28 Policy Implications: Results All policies except mandated sense-and avoid hardware had little effect on the cost advantage of UAS over manned alternative Required sense-and-avoid hardware greatly affects cost-effectiveness, however Localized Surveillance and Pipeline Monitoring would no longer be viable as larger, much more expensive UAS would be necessary 29 Risk, Technologies, & Standards Team Members: Samiah Akhtar Jonathan Cornell Nicole Hayward Will Kim Nick Misek Doug Robl Team Manager: Keith Florig 30 Purpose Derive a risk model to explore how risk is related to UAS numbers, dimensions, and flight zones Research on elements of risk mitigation such as human factors, sense and avoid Exploration of alternative incident reporting systems Predator Source: http://www.fs.fed.us/psw/news/PSW_ News/2005_09/images/uav.gif 31 Technology and Risk Outline Goals Risk Modeling Purpose Assumptions & Approach Findings UAS Risk Mitigation 32 Risk Modeling Purpose Provide a way of modeling that creates some groundwork for future modeling Use model to compare relative risk calculations Pointer to the future, not the answer Points of interest Mid-air vs. single-craft crash Effect of sense and avoid technology UAS to displace manned aircraft Source: http://www.maximog.com/ima ges/sublevel/uav_left.jpg 33 Risk Modeling Assumptions Uniform national model Uniform traffic density Uniform ground population density Uniform aircraft per type Appropriate for: VFR traffic Rural, less populated areas NOT Appropriate for: Urban settings Airports High traffic densities 34 Risk Model 35 Risk Modeling Approach Number of midair collisions: N [1 P( A)] D 2 S 31 10 6 sec/ year N = total number of aircraft in defined airspace ρ = aircraft traffic density D = diameter of plane (wingspan) S = average aircraft speed P(A) = probability of avoidance (Used for calibration) VFR operations only UAV Picture Source: http://www.evworld.com/press/spider_lion_uav.jpg 36 Risk Modeling: UAs displacing Manned Expected Annual Fatalities vs. % of Airspace Unmanned Expected Annual Mortality 14 12 10 8 6 4 2 0 0% Small risk from 10% 20% unmanned at lower Risk from extrema unmanned at low levels less than decreased risk from manned 30% 40% 50% 60% 70% 80% 90% 100% % of Total Aircraft is Unmanned Single-craft crashes still present less risk than midairs Manned Mid-Airs Manned Single-Craft Unmanned Mid-Airs P(Avoid)=0 Unmanned Mid-Airs P(Avoid)=0.5 Unmanned Single-Craft 100 failures/100k flight hrs Unmanned SIngle-Craft 10 failures/100k flight hrs Unmanned Single-Craft 500 failures/100k flight hrs 37 Risk Modeling: Mid-Air vs. Single-Craft Expected Annual Mortality vs. % UASs Added to NAS Expected Annual Mortality 40 35 30 At some point, manned risk surpasses unmanned risk 25 20 15 10 5 0 0% 20% At low numbers, sense and avoid has little effect 40% 60% 80% 100% 120% % UASs Added 140% Single-Craft 50 failures/100k flight hrs Single-Craft 200 failures/100k flight hrs Mid-Airs P(Avoid)=0 Mid-Airs P(Avoid)=0.25 Mid-Airs P(Avoid)=0.5 Mid-Airs P(Avoid)=0.95 Single-Craft From Substituting Manned Mid-air from Substituting Manned 160% 180% 200% Single-Craft generally less risk than mid-air 38 Risk Modeling Conclusions Mid-air collisions generally have more risk than single-craft crashes Displacing small to moderate amounts of manned craft represents decrease in risk Smaller, less reliable UAs can present less risk than larger more reliable manned aircraft For small numbers of UAs in low traffic densities, sense and avoid has small effect 39 Technology and Risk Outline Goals Risk Modeling UAS Risk Mitigation Human Factors Sense and Avoid 40 Human Factors Implications Risks - Caused Most Number of Accidents “Sensory Isolation” [McCarley et al] UAS operator does not receive same sensory cues as manned aircraft operator Automation Malfunction of automated components controlled by the UAS operator Operator Hand-Off Issues with handing off control of vehicle from one operator or crew to another 41 Human Factors Implications Recommendations Training and Procedures Up to date training as new technology advancements arise Ensure that operator has accurate knowledge of automated components within UAS Multimodal displays Prevent sensory isolation Allow for audio, visual and speech control Example: simulated cockpit 42 Detect, Sense and Avoid Risks Market impact of single fatal collision Lack of standardization among DSA systems Sense and Avoid Technology Cost* Size* Weight* Power Usage* Pros Cons IFF Transponder $500-3000 1.5m2 6 kg 30 watts, nominal Low-cost, Transponder requirement ADS-B Transponder $2000-6000 2.0m2 8 kg 50 watts, nominal Compatible with NGATS High power, transponder EO/IR Sensors $50,000 200,000 4.0m2 5-10 kg 10-20 watts Detection in IFR and VFR, no metal requirement Short detection range, high cost Synthetic Aperture Radar $10,000 50,000 3.0m2 20 kg 80 W All-weather Weight, power - high 43 Detect, Sense and Avoid Recommendations Create regulations specific to size, weight, application etc Testing Periods Phased Integration 44 Technology and Risk Outline Issues Goals Risk Modeling UAS Risk Mitigation Reporting systems 45 Current Reporting Systems Two Options NTSB Reporting (as required by FAR) - Accident NASA ASRS Voluntary Reporting - Incident Current Implementation NTSB mandates detailed information when: Flight control system malfunction, Illness of crewmember, Turbine Engine Failure, Inflight fire, Mid-air collision or Damage in excess of $25,000 to other property ASRS System is anonymous and does not have any reporting requirements 46 Reporting Recommendations Initially mandate reporting of all accidents and incidents Re-evaluate strategy after testing period NTSB - NTSB information helps FAA to assess standards - NTSB provides useful information on UAS failures - FAA responds with rules for reporting incidents. - UAS responds with improved design and engineering Communication Triangle FAA UAS manufacturers 47 Public Awareness & Perceptions Team Members: Darian Ghorbi Jenny Kim Mark Peterson Laura Seitz Patrick Snyder Team Manager: Pete Tengtrakul 48 Statement of Purpose Add the element of public perception to the discussions of UAS in the NAS Motivation: the fact that there has never been a formal presentation of public perception on the topic Findings: useful for the creation of regulations and policy implications 49 Objectives Compare public perceptions of the risks concerning manned and unmanned aircraft Find demographic groups with certain risk and benefit patterns of UAs Research implications of opinion of UAs Create survey to aid in completing objectives 50 Hypotheses Perceived Risk - Manned < Remotely Piloted < Autonomous Ground vs. Air - More risk of UAS perceived in air Prior Knowledge vs. Risk Perception - Prior knowledge, associate less risk Benefit vs. Risk Perception - Higher benefit, lower risk Education vs. Risk Perception - Technical education, associate less risk Age vs. Risk Perception - Older participants more cautious Frequency of Flight - Those that fly frequently, associate less risk 51 Layout of Survey First Page Provide information about UAS Autonomous Remotely Piloted Gauge previous knowledge Source Last Page Demographics Gender Age Education Frequency of Flight Voting (identify opinions of those that are politically engaged) Pilot 52 Layout of Survey Application •Traffic Monitoring •Pipeline Monitoring •Disaster Relief •Border Patrol Questions • Quick Response • Benefit • Picture of UAS application • Description of UAS •Physical Information • Stakeholder • Public Risk • Ground • Air 7 Point Scale • 1 - Much Less • 4 - Same • 7 - Much More •Current application 53 Obtaining Surveys Method Mon Valley NGO Word of Mouth Total # of Surveys 79 (56%) 62 (44%) 141 Coding: numerical code assigned Screening: data obtained from those under 16 years of age were not counted 54 Statistical Methods Paired T-tests Across applications ANOVA Significance of mean Regression Correlations Demonstrated the strength of the variables (risk and benefit) 55 Results: Descriptive Statistics Demographic Variable Mean Scale Previous Knowledge Gender Age Educa tion Register to Vote Freque ncy of Fli ght Licensed Pil ot 0.53 0.44 36.04 2.33 0.66 2.51 0,01 1= heard, 0 =never heard 1=Male, 0=Female Years 1=High School,4= Gradua te 1=Registered,0=not 1=never, 5=more than 12x 1=Licensed,0=not Variable Remotely Piloted Autonomous Support .75 (0.45) 4.65 (1.57) 4.14 (1.58) 4.36 (1.51) 4.55 (1.39) .58 (0.50) 4.29 (1.77) 4.01 (1.66) 4.81 (1.66) 4.87 (1.68) Benefit to Stakeholders Benefit to Society Risk to people on Ground Risk to people in the Air 56 Perceived Risk (1=Low, 7=High) Perceived Relative Risks Between Remotely Operated vs. Autonomous 7 6 5 4 Pipeline T raffic Border Disaster 3 2 1 Remotely operated UA Autonomous UA 0 UA Applications Autonomous applications are viewed to have more risk in comparison to remotely operated UAs. 57 Relative Risks Across Applications Perceived Risk (1=Low, 7=High) 7 6 5 4 3 Pipeline Traffic Border Disaster 2 1 Remotely operated UA Autonomous UA 0 UA Applications Traffic Monitoring has the highest perceived risk. 58 Perceived Benefit (1=Low, 7=High) Relative Benefit Across Applications 7 6 5 4 3 Pipeline* Border Disaster T raffic 2 1 Remotely operated UA Autonomous UA 0 UA Applications The more risky the public perceived the application, the less benefit they associated with the application. 59 Relative Perceived Risk to People on Ground vs. Air Perceived Risk (1=Low, 7=High) 7 6 5 4 Pipeline* Traffic* Border* Disaster* 3 2 1 Risk to people on Ground Risk to people in the Air 0 UA Applications There is no difference between risk perceived on ground vs. air. Also, there is no difference between perceived benefit between stakeholders and society. 60 Demographics and Risk Age N Mean StDev Under 25 56 4.340 1.541 36-35 28 4.679 1.786 36-45 13 5.192 1.665 46-55 14 4.964 1.216 56-65 21 5.310 1.512 Over 65 9 2.333 1.871 Those over the age of 65 perceived UAs as least risky and least beneficial; the mean value is insignificant. 61 Risk Perception Conclusions Unmanned aircraft risk > manned Autonomous risk > Remotely piloted No difference: Risk: Ground vs. Air Benefit: Stakeholders vs. Society 54% heard of UAS Need education programs 78% of those that heard of UAS obtained information from television The more familiar, the more comfortable Traffic Monitoring - higher risk Fear of operating around high population density areas 62 Impact on Policy Limit flight path/area Limited population density Implement education/outreach programs 63 Future Insights Limitations: Time, Resources, and Budget Sample National scale-different regions Future Surveys: Compare UAS to other risky technologies Size of aircraft Privacy concerns Economics concerns Lengthened 64 Governance Team Members: Nora Darcher Norma Espinosa Scott Fortune Andrea Fuller Team Manager: Leonardo Reyes-Gonzalez 65 Purpose Evaluate current system of governance for UAS integration against principles of good governance Suggest measures that could improve the governance process 66 Analysis Principles of Good Governance Rules for FAA governance Historical Technologies Actor Interactions Roadblocks Cost and Benefits for each actor 67 Characteristics of Good Governance http://www.unescap.org/pdd/prs/ProjectActivities/Ongoing/gg/governance.asp 68 Governance requirements on FAA OMB rule requires FAA standards adoption procedures to have the following: Openness Balance of interest Due process Appeal process Consensus 69 Historical Analysis How did the governance system handle the introduction of new technologies? Technology What We Learned Supersonic Transport Public Interest NGOs can have a large impact (1960s) Automation of Radar System (1970s) Microwave landing system (1970s) Incremental changes are easier than changing entire system at once Intl. adoption of US standard is advantageous to US technology firms 70 Actor Analysis Objective: Provide a systematic assessment of the actors involved in integrating UAs into the NAS Process: Identified key actors, examined their goals and looked at problem from each actor’s perspective 71 Actor Analysis Resistant Neutral For Hesitant 72 Roadblock Analysis Objective: prioritize problems inhibiting the integration of UAs Categories Technological Organizational Infrastructural Public Concern 73 Roadblock Analysis 74 Roadblock Analysis # of actors 6 Airspace Access 5 4 Equivalent/Acceptable Level of Safety 3 2 1 Transponders Data Acquisition 0 Low High Complexity 75 Most obvious needs Defining an equivalent/acceptable level of safety Allowing UAS operations in scarcely-used airspace to facilitate testing and development for civil and commercial applications. Potentially large public concern about UAS safety argues for proactive public involvement in deliberations 76 Conclusions 77 Summary of Conclusions Economics Risk, Technologies, and Standards Public Awareness and Perception Governance 78 Economics Some civil UAS applications seem highly competitive with alternatives Initial policy ought to be tailored to the most commercially viable applications Cost models show that (i) costs are most sensitive to hours of utilization, (ii) safety equipment has modest cost effect, except for small systems using sense and avoid Foreign UAS firms may develop an advantage if they gain airspace first 79 Risk, Technologies, and Standards For some applications in some classes of airspace, unmanned aircraft result in fewer fatalities than manned aircraft used for the same task Sense and avoid is important only in airspace with significant traffic density Low risk areas could be used for experimentation and testing without posing a high risk to those on the ground or in other aircraft A mandatory incident reporting system has potential to greatly improve both airworthiness and human factors reliability 80 Public Awareness and Perception All UAS applications surveyed were considered more risky and less beneficial than the manned alternative Traffic monitoring perceived as most risky (likely due to flight over dense population) About half of participants had heard of UAs Those more familiar with UAS technology perceive less risk 81 Governance Integration problem is more complex than many people realize Incremental approach allows for policy experimentation at low risk (e.g., sparsely populated areas/airspace) Standards need to be established to provide benchmark and incentive for manufacturing Attention to public perception and involvement can greatly influence unfolding of UAS issue 82 Thank You for Coming! Questions? 83