SOFT COMPUTING IN THE SMART COCKPIT A Workshop The 2003 International Symposium on Intelligent Control Houston, Texas October 5-8, 2003 ISIC-2003-1 Valasek, Ioerger, Painter INSTRUCTORS Dr. John Valasek - Assoc. Professor, Aero. Eng., Texas A&M. valasek@aero.tamu.edu 979 845-1685. Dr. Tom Ioerger - Assoc. Professor, Cptr. Sci., Texas A&M. ioerger@cs.tamu.edu 979 845-0161. Dr. John Painter - Professor, EE, Aero., Cptr. Sci. (Ret.), TAMU altair@tca.net ISIC-2003-2 979 696-0429. Valasek, Ioerger, Painter COURSE OUTLINE SECTION I • Introduction To The Smart Cockpit And Free Flight • Free Flight Functionality • Architecture: Aircraft & Simulator, Software & Hardware SECTION II • Artificial Intelligence In The Cockpit • Intelligent Computing Techniques • Software Agents: The Key To Smart Cockpit Software SECTION III • Real Time Flight Simulation: The Basics • Real Time Flight Simulation For Free Flight SECTION IV • Selected Research Results • Conclusion: Summing It Up ISIC-2003-3 Valasek, Ioerger, Painter SMART COCKPIT COMPUTING - WHAT IS IT? • COCKPIT FUNCTIONAL INTEGRATION VIA SOFTWARE Helping the Pilot Visualize, Understand, and Fly. Helping the Airplane Automate Nominal Flying Tasks. Helping Air Traffic Control With Transition to “Free Flight.” • VIEWING AVIATION AS A SOFTWARE PROBLEM. Reducing Pilot Task Load Via Automation. Increasing General Aviation (GA) Pilot Performance in Weather. Increasing GA Weather Access to Un-Instrumented Airports. Increasing Flight Safety With - Automatic Collision Avoidance. - Automatic Weather Avoidance. - Automatic Terrain Avoidance. - All Three at the Same Time. ISIC-2003-4 Valasek, Ioerger, Painter SMART COCKPIT COMPUTING - WHY IS IT? • HAVE YOU FLOWN, LATELY? Major City Airports are Traffic Saturated. Air Travel is Geared to Using Major Cities. Only 715 Airports are Weather Instrumented. The Answer is Not More Jumbo Jets. • WHY NOT FLIGHT-ENABLE THE MEDICALLY QUALIFIED? There are 5,400 Public Airports. Flying Can Become a Public Utility. Computing Technology Can Satisfy the Major Requirements. - Inexpensive Airport Weather Instrumentation . - GA Aircraft Relative Affordability. - GA Pilot Weather Proficiency Increased. - GA Pilot Training Affordability. ISIC-2003-5 Valasek, Ioerger, Painter IS THE SMART COCKPIT JUST FOR GA? • TECHNIQUES APPLY TO AIR TRANSPORT AS WELL AS GA. Improves Situational Awareness and Multi-Tasking. Reduces Pilot Work-Load. Works for the Pilot, Not Vice-Versa. Works for the Pilot, But Doesn’t Replace Him. • BUY THIS TECHNOLOGY “BY THE YARD.” Software Architecture is Modular - by Function. Some Functions Require Additional Cockpit Instrumentation. Implement Functions as You Can Afford Them. Augmentable Functionality is the “Name of the Game.” ISIC-2003-6 Valasek, Ioerger, Painter SMART COCKPIT COMPUTING - WHERE IS IT? • AT THE CROSS-ROADS OF MULTI-DISCIPLINES. Flight Control & Air Traffic Management. Intelligent Control & Soft Computing. AI & Software Agents. Human Factors & User Interface Design. • IN THE HANDS OF MULTIPLE AGENCIES. Government - NASA Langley Research Center (SATS). Academia - Aeronautical Engineering, Computer Science, Etc. Industry - Avionics Companies (FMS, GPS, etc.). ISIC-2003-7 Valasek, Ioerger, Painter SATS - THE PROGRAM Small Aircraft Transportation System • FIVE-YEAR DEMONSTRATION GOALS (2001-2005) Technology and Operational Capability: 1. Higher Volume @ Non-ILS Airports 2. Decreased Landing Minimums (Weather). 3. Increased Single-Pilot/Mission Safety/Reliability. 4. Integration of SATS Traffic With NAS. • THE OPERATIONAL/TECHNOLOGY KEYS. Increasing Cockpit Technology Increases Operational Capability. “Self-Controlled Airspace” (SCA) for Smaller Airports. Increasing “Usable” Airports by 750% (4,685 + 715 = 5,400). ATC-Acceptable Procedural/Spatial Interface and Hand-off. ISIC-2003-8 Valasek, Ioerger, Painter SATS - THE APPROACH Small Aircraft Transportation System ATC: FAA Air Traffic Control. IAF & FAF: Initial- and Final-Approach Fixes. ADS-B: Automatic Dependent Surveillance Broadcast (Radar Xpndr.) FAF RUNWAY AMM: Airport Management Module (Digital Data-Link) • ATC Clears Aircraft to SCA Holding Stack at IAF. • Self-Separation via ADS-B (Req. Conflict Mgt. Software). • Approach Sequencing and Airport Info. via AMM. ISIC-2003-9 Valasek, Ioerger, Painter SMART COCKPIT RESEARCH - DOING IT • REQUIRES A FIXED-BASE FLIGHT SIMULATOR. General-Purpose Cockpit. Realistic Controls. - Stick, Rudder, Throttles, Flaps, Trim, Gear, Brakes. - It Must be Realistic to Fly. General Purpose Panel Displays. - Touch-Screen LCDs Are Good (for Research/Development). Forward-Projection Screen System. - At Least 90-degrees Wide. - 160-degrees is Better (Wrap-Around, 3-Screen). • THE SIMULATOR IS A SOFTWARE DEVELOPMENT TOOL. Software Modules Developed in MATLAB® - Ported to C++. Software Modules Integrated in the Flight Simulator Computers. Software Functionality Validated/Corrected by Pilots Flying It. ISIC-2003-10 Valasek, Ioerger, Painter FREE FLIGHT - WHAT IS IT? • A MOVING TARGET. Its Specification is Not Yet Stable. Ideal Agreed, but Details Disagreed. Players: FAA, NASA, Aviation Industry. • THE ORIGINAL IDEAL1 “Freedom of Choice” … of IFR Routes (RTCA - 1995). Pilot Selection of Trajectory … in Real Time. “Max/Max” Solution for Safety/Efficiency. • THE PROBLEM. Requires New Air/Ground Operations and Technology. 1Control Applications and Challenges in Air Traffic Management, by Joseph W. Jackson and Steven M. Green, Proceedings of the American Control Conference, June, 1998, pp. 1772-1788. ISIC-2003-11 Valasek, Ioerger, Painter FREE-FLIGHT - COCKPIT FUNCTIONALITY • VOICE RADIO. • NAVIGATION • PANEL INSTRUMENTS • SITUATION DISPLAY - Global Positioning System (GPS). - Flat LCD With Touch-Screen. - Flat LCD With Touch-Screen. (Moving Map Reqs. Nav. Data Base). • AVOIDANCE FUNCTIONS: Collision - On Situation Display (Reqs. ADS-B Beacon Add-on). Weather On Situation Display (Reqs. A/G Digital Data Link). Terrain - On Situation Display (Reqs. Nav. Data Base). • HEAD-UP DISPLAY (HUD) - For Eyes Out of the Cockpit. - Instrument and Approach Displays. • APPROACH AIDS: - GPS, ILS, and/or SATS. - On HUD and/or Situation Display. • PILOT ADVISOR - On HUD and/or Situation Display. (Req. Flight Mode Interpreter) ISIC-2003-12 Valasek, Ioerger, Painter BASIC GUIDANCE LOOP(S) FLIGHT MANAGEMENT SYSTEM FLIGHT PLAN Closed-Loop Open-Loop NAVIGATION + GUIDANCE AUTOPILOT PILOT Manual CONTROL AIRFRAME NOTE: CDU Interfaces Not Explicit ISIC-2003-13 Valasek, Ioerger, Painter THE BASIC NAVIGATION TRIANGLE The Problem is “Wind” Air Vector: Heading/True-Air-Speed = 090°/150 Wind Vector: WD/WV = 350°/20 Ground Vector: Track/Ground-Speed = 097°/155 • CONVENTIONS. Map Convention: North (N) is “Up.” Directions Measured Clockwise, From True North (360°) Air Vector and Ground Vector are “To” Direction. Wind Vector is “From” Direction. • AIR NAVIGATION REQUIRES COMPUTATION OF “WIND.” Manual Wind Computation Uses Mechanical Computer - “E6B.” Automated Digital Wind Computation in Avionics. ISIC-2003-14 Valasek, Ioerger, Painter GLOBAL POSITIONING SYSTEM (GPS) SV1 • THE NEW SOLE MEANS OF AIR NAVIGATION. SV9 • A SATELLITE MULTI-RANGING SYSTEM Global Coverage. All-Weather Capability. 24 12-Hour Satellites (3 On-Orbit Spares). Position/Velocity Output Latitude, Longitude, & Altitude. 4 Range Measurements - 3 Position Coordinates SV17 & Precise Time (12 pico-sec.). Position Error: ~ A Few Meters. Position/Velocity Output - Lat., Long., & Alt. Soon to Replace Visual Omni Range (VOR) & Instrument Ldg. System (ILS). ISIC-2003-15 Valasek, Ioerger, Painter NAVIGATION/GUIDANCE FUNCTION Manual Flying • BASIC NAVIGATION FUNCTION. GPS - Time, Position (LAT, LNG, ALT), Velocity (TRK, GS, ROC). Wind Computing - A/C Data Input (MAG HDG/VAR, IAS, TEMP). Guidance (Pilot Computed) - Waypoint HDG, ETE, ETA. •AUGMENTED NAVIGATION FUNCTON. Navigation Data Base (Waypoint Location Data). Guidance Computing - Adds Altitude (AGL) and Approach NAV. • HIGH-LEVEL NAVIGATION FUNCTION. Flight Plan Driven - Requires Pilot Entry of Flight Plan. Guidance Computing - Adds Waypoint Maneuver Alerts (Turn, Climb). ISIC-2003-16 Valasek, Ioerger, Painter NAVIGATION/GUIDANCE FUNCTION Autopilot Flying • AUTOPILOT (A/P) FUNCTION. Autopilot Drives A/C Control Surfaces (Servo). - Ailerons, Elevator, Rudder, Throttle (Optional). Autopilot Inputs - HDG, ALT, IAS (Optional). Inputs Manually by Pilot, or Computer-Generated. •FLIGHT MANAGEMENT SYSTEM (FMS) FUNCTION. Automates Guidance for High-Level NAV. Computes and Issues Guidance Commands to Autopilot. Simultaneous Automatic Navigation and Maneuver. Automates All Flight Phases, Including Approach/Landing. • NAVIGATION/GUIDANCE FUNCTIONAL HIERARCHY. Manual Flying Autopilot FMS ISIC-2003-17 Valasek, Ioerger, Painter COCKPIT DISPLAYS, BY FUNCTION • THE PILOT’S VIRTUAL WORLD. It’s All About Situational Awareness. The Pilot “Reckons” His Situation in 4D Space-Time. Build a 4D Mental Image … Using 2D Displays. • DISPLAY DIFFERENTIATION AND TYPING. “Internal” Situation - The Airplane (Trad. Gauges & Switches “External” Situation - The Flight. Long-Term - Relatively Static (Trad. Maps & Books). Short-Term - Very Dynamic (Traditionally, Gauges). Internal and Short-Term External Display Commonalities. Two Different Genre of Display. ISIC-2003-18 Valasek, Ioerger, Painter COCKPIT DISPLAY - GUIDANCE & NAV • DISPLAYS TAILORED TO TEMPORAL INFO I/O NEEDS. • INFORMATION OUT, ONLY. Long-Term - Navigation (ex. - Moving Map on Panel Display). Short-Term - Guidance/Control (ex. - A/C Dynamics on HUD). Continuously Functioning, Disparate Displays. • INFO INPUT AND OUTPUT - Flight and/or System Management. Long-Term - Flight Planning (ex. - One mode of Panel MFD). Short-Term - System Management (ex. - A/P, FMS Modes). Sequentially Functioning, Multi-Function Display (MFD). I/O Displays Called “Control/Display Unit” (CDU) ISIC-2003-19 Valasek, Ioerger, Painter THE VIRTUAL RUNWAY • HUD DISPLAY OF A “VIRTUAL” RUNWAY OUTLINE. Generated from GPS and Aeronautical Data-Base. “Breaking Out” at 200 Feet, on an ILS Approach. ISIC-2003-20 Valasek, Ioerger, Painter HAZARD AVOIDANCE FUNCTION • HAZARDS TO FLIGHT. Severe Weather - (ex. - Thunderstorms). Conflicting Air Traffic - (ex. - VFR Collisions). Controlled Flight Into Terrain - (ex. - Mountain Flying). • FREE FLIGHT HAZARD AVOIDANCE REQUIREMENTS. Cockpit Acquisition of Hazard Data - (Digital Radio Data). Cockpit Computation of Avoidance Trajectories - (Guidance). Cockpit Display of Hazard/Avoidance Imagery - (Map). • FINER POINTS. Cockpit Trajectory-Balancing Between Multiple Hazards. Automated Negotiation Between Aircraft and Air Traffic Control. ISIC-2003-21 Valasek, Ioerger, Painter COCKPIT DISPLAY - HAZARD AVOIDANCE • THREE HAZARDS OF DIFFERING TEMPORALITY. Terrain - Long-Term (No Dynamics). Weather - Long- to Medium-Term - (Dynamics Not A/C-Scale). Traffic - Medium- to Short-Term - (Dynamics A/C-Scale). • DISPLAY CHOICES. All Three Situations Require Info-Out, Only, Display. Imagery of All Three Hazards Can Overlay Moving Map (NAV). Long-Term Guidance Vectors Can Also Overlay Moving Map. Display “Clutter” is a Human Factors Issue, Here. Short-Term Steering Commands on A/C Dynamics Display (HUD). ISIC-2003-22 Valasek, Ioerger, Painter SOFTWARE ARCHITECTURE Form Follows Function • BUILDING A DEVELOPMENT TOOL AND ENVIRONMENT. When You’re Up to Your Armpits in Alligators … ? … Remember, the Goal is a Research Tool. Don’t be a Slave to Current Cockpit Avionics. This Problem is Like GPS … 90% Software. • HOW TO LAY OUT THE SOFTWARE ARCHITECTURE. Modularize, Modularize, Modularize !! Structure the Whole, Function by Function. Go for Independent, Communicating, Software Modules. - “Independence” as in Task Independence, not Data Independence. - Hence, “Communicating” Modules (AI’s “Message-Passing”). Let the Computing Hardware Take Care of Itself. Think About Project Management and Configuration Control. ISIC-2003-23 Valasek, Ioerger, Painter ARCHITECTURE - DATA-FLOW DIAGRAM Aircraft Dynamics HUD FCS Pilot A/P CDU FMS FLT PLN NAV NAV DB MFD TFC GEN (MAP) WX GEN TFC AVD EXEC AVD WX AVD ISIC-2003-24 Valasek, Ioerger, Painter SOFTWARE ARCHITECTURE REALIZATION IN A FLIGHT SIMULATOR • COST-EFFECTIVE SIMULATOR IMPLIES … … DISTRIBUTED COMPUTING. Use “Simple” Computers, One Per Display. Displays: - MFD, CDU, HUD, and the Three-Screen, Projected “World.” Projecting “The World” May Not Be So Simple - Graphics Engine. - Surface Geometry Generation is Compute-Intensive. - Pilot Requires 30 Frame per Second Refresh Rate. Use a “Projected HUD” - Simplify Cockpit Hardware. Simulator Controller Station - One More Computer & Display. Settle on 4 Computers - 3 MS-Windows PCs, 1 SGI Unix Machine. ISIC-2003-25 Valasek, Ioerger, Painter SIMULATOR HARDWARE ARCHITECTURE Projection Screen With HUD Projector Center CPTR GRAPHICS Flaps Throttles Stick & Rudder Gear & Trim MFD (button) CDU (touch) CTRLR (kybd) PC-1 MFD PC-2 CDU PC-3 CTRLR Ethernet Multiplexed Serial Port Adapter ISIC-2003-26 Valasek, Ioerger, Painter ARTIFICIAL INTELLIGENCE IN THE COCKPIT • A FUZZY “FLIGHT MODE INTERPRETER.”1 Fuzzy Decision Tool - Bayes Connectives. Flight “Modes” as State Partition. Taxi; Take-off; Climb-out; Cruise; Hold; Initial, Final, and Missed Approach; and Land - 9 Modes. Membership Functions Modeled for Particular Aircraft. • A RULE-BASED “PILOT ADVISOR.” Keeping the Flight Within the “Envelope.” Mode-based - Driven by Flight Mode Interpreter. Rules for Instrument Flight. Rules for Performance of This Particular Airplane. 1 “Hypertrapezoidal Fuzzy Dynamic State Interpreter,” U.S. Patent 6,272,477,B1, by Wallace E. Kelly, III and John H. Painter, Aug. 7th, 2001. ISIC-2003-27 Valasek, Ioerger, Painter HYPERTRAPEZOIDAL MEMBERSHIP FUNCTIONS • A 2-DIMENSIONAL PROJECTION OF A 9-DIMENSIONAL M.F. 9 State-Variables and 9 Modeled Flight Modes. CRUISE INITIAL APPROACH ` 1 0.5 FINAL APPROACH 0 150 LANDING 3000 100 airspeed [knots] 2000 50 1000 0 0 ISIC-2003-28 altitude [feet] Valasek, Ioerger, Painter FUZZY LOGIC - OTHER COCKPIT APPLICATIONS • AUTOMATING DISPLAY CALL-UP AND FORMATTING. Mode-Driven, With Pilot Over-ride. • BLENDING MULTIPLE GUIDANCE TRAJECTORIES. (A Fuzzy Executive Guidance Agent.) Multiple Hazard-Avoidance Trajectories. Example: Nominal Flight Plan versus Weather and/or Traffic. Define Spatial Trajectory-Risk Functions. Normalize the Set of Evaluated Risks (0-1). Prioritize Individual Avoidance Trajectory Generators. Minimize Highest Risk, First, Then Lower Risks - Sequence. ISIC-2003-29 Valasek, Ioerger, Painter COURSE OUTLINE SECTION I • Introduction To The Smart Cockpit And Free Flight • Free Flight Functionality • Architecture: Aircraft & Simulator, Software & Hardware SECTION II • Artificial Intelligence In The Cockpit • Intelligent Computing Techniques • Software Agents: The Key To Smart Cockpit Software SECTION III • Real Time Flight Simulation: The Basics • Real Time Flight Simulation For Free Flight SECTION IV • Selected Research Results • Conclusion: Summing It Up ISIC-2003-30 Valasek, Ioerger, Painter Intelligent Computing Techniques Software Engineering Object-oriented programming Data communications: TCP/IP Modular design Message Passing Intelligent Agents model complex decision-making/protocols simulate autonomous behavior ISIC-2003-31 Valasek, Ioerger, Painter Data Communications TCP/IP enables modularization & scalability (>1 machine) example: connect independent flight dynamics engines to common server typically 10MB/s pass text strings (e.g. set/get state vars) buffering, blocking ISIC-2003-32 Valasek, Ioerger, Painter 128.135.194.2 128.135.194.3 SERVER Create socket on port 5000 CLIENT listen for connections Open Socket connection to 128.135.194.2, port 5000 accept connection write block of bytes read block of bytes write block of bytes ISIC-2003-33 read block of bytes Valasek, Ioerger, Painter Two Flavors: TCP vs. UDP specify type when create socket TCP guaranteed delivery; messages arrive in order checked for errors UDP non-guaranteed; not error-corrected much faster! choice depends on tolerance of msg failures example: comm. protocol vs. graphics updates ISIC-2003-34 Valasek, Ioerger, Painter Inter-Module Communications FLT SIM PA TCP/IP TCP/IP CMD GEN HUD GEN A/P DSPLY HNDLR FCS FLT CTRLS EQMO DSPLY HNDLR FMI COCKPIT DATA SIM SIM RADAR SIM ADS-B NAV MOD WX AGT TFC AGT FLT PLN FMS EXEC DATA OBJ SIM CPDL FMS LOGIC HDD Left PILOT JPSN DB Dashed lines denote virtual connectivity. HDD Right HUD DSPLY HNDLR SPiFI ISIC-2003-35 Valasek, Ioerger, Painter Other Communications Tech. CORBA - industry standard www.omg.org/gettingstarted/corbafaq.html network/object-oriented, location-independence Interface Definition Language: classes, methods remote method invocation (lang/OS independent) HLA - High Level Architecture www.dmso.mil/public/transition/hla standard for mil/gov simulations federations, broadcasts, time management objects, interactions SOM/FOM - interface definitions, methods ISIC-2003-36 Valasek, Ioerger, Painter XML Data representation format for scenarios, msgs/cmds, config, logs, etc. Define tags (like HTML) <events> <simevent id=“54321” time=“1:21”> <type>request_takeoff_clearance</type> <aircraft>US789<aircraft> <location>KDEN</location> </simevent> <simevent> ... </simevent> </events> ISIC-2003-37 Valasek, Ioerger, Painter Generic parsers available Xerces - Java, C++ Apache: xml.apache.org/xerces2-j IBM: www.alphaworks.ibm.com/tech/xml4j SAX => incremental, parse when needed DOM => batch, produce “object trees” root Doc node (list of child nodes...) Element: type text=takeoff Element: simevent attribute: time=1:21 attribute: id=54321 Element: location text=KDEN ISIC-2003-38 Element: simevent attribute: time=1:35 attribute: id=54322 Element: aircraft texst=US789 Valasek, Ioerger, Painter What are Agents? Essential Characteristics: Situated can sense and take actions in dynamic environment Goal-oriented Autonomous Social/collaborative Adaptive ISIC-2003-39 Valasek, Ioerger, Painter Agent Architectures Production Systems Reactive, trigger rules, CLIPS, SOAR Search Algorithms: A* (WX agent) Planning Algorithms Hierarchical Task Networks (Retsina, TRL) Decision Theoretic Markov Decision Processes, maximize payoff Cognitive (Mentalistic) BDI: beliefs, desires, intentions JACK, PRS, dMARS ISIC-2003-40 Valasek, Ioerger, Painter Roles for Agents in Aviation Simulate other aircraft, controllers In cockpit: planning flight path, managing fuel, maintaining stability of flight, monitoring traffic or weather conflicts… On ground (TRACON, ARTCC): planning trajectories, resolving conflicts, approach metering, handling emergencies, coordination with ground ops, airlines, etc. ISIC-2003-41 Valasek, Ioerger, Painter Collaboration Models Teamwork Hierarchical vs. distributed (command vs. consensus) Key concepts: roles and responsibilities Shared plans: implicit coordination, synchronization Theoretical basis: Joint Intentions Negotiation protocols Distributed Constraint Satisfaction Share justifications and beliefs to determine compromise Monotonic Concession Protocol Auctions Bids based on marginal utility Contract networks ISIC-2003-42 Valasek, Ioerger, Painter Role of Simulated “Mental Attitudes” Intent – transmit more than position/vector Desire to avoid weather, flight plan, will be turning north, descending due to turbulence, reason for deviation… Beliefs shared info (weather, congestion, aircraft emergencies) common picture of situation common knowledge: STAR’s, fixes, active runways, traffic patterns manage uncertainty ISIC-2003-43 Valasek, Ioerger, Painter Concepts for Development of MultiAgents for Free Flight Strategic (trajectory planning/management) vs. Tactical (avoidance maneuvers) Actionable decisions: Alter flight path: heading, altitude, speed Factors: weather, terrain, traffic Constraints: fuel, speed/alt range Preferences: time, fuel cost, comfort ISIC-2003-44 Valasek, Ioerger, Painter Negotiation Utility function: Flight Plans => score Negotiation by “argumentation” State what is wrong with proposed solution and why Communicate preferences as well as constraints make up when behind schedule minimize fuel consumption maneuver limitations (safety, comfort) Monotonic Concession Protocol (Rosenschein and Zlotkin) define a finite set of alternative trajectories each agent ranks trajectories by utility, proposes best take turns proposing next best deal till utilities match ISIC-2003-45 Valasek, Ioerger, Painter Negotiation Flowchart no weather, terrain, traffic info. identify & contact other aircraft (bogey) conflict detection generate alternative trajectories receive counterproposal from bogey propose next-best solution for ownship rank them by utility is utility of bogey’s no solution>=ownship’s yes are there alternatives left? conflict detected? yes propose highest-ranked solution for ownship yes exchange candidate solutions with bogey confirm agreement on last proposal no ask ATC for help inform ATC ISIC-2003-46 Valasek, Ioerger, Painter TRL Agents “Task Representation Language” Developed at Texas A&M Comp. Sci. Dept. contact: ioerger@cs.tamu.edu knowledge bases declarative: rule base (“domain knowledge”) procedural: plans/methods for achieving goals connection to simulator read state information trigger actions agents can communicate with each other ISIC-2003-47 Valasek, Ioerger, Painter TRL Agent Architecture TRL agent TRL Task Decomposition Hierarchy TRL KB: tasks & methods APTE Algorithm Process Nets operators messages Simulation ISIC-2003-48 assert, query, retract results sensing JARE KB: facts & Horn-clauses messages Other Agents Valasek, Ioerger, Painter Example Task Description (task flight-plan-1 () (method (sequence (takeoff KCLL 16) (climb-out 3000) (turn-heading 350) (fly-direct-to KCNW) (descend 500) (land KCNW 17L)))) command to simulator invoke sub-task Things to add: • interaction with ATC (set new way-points, altitudes...) • handling developing weather (while (not cloudy)...) ISIC-2003-49 Valasek, Ioerger, Painter The SATS Airport Controller How to simulate this with agents? encode the formal protocol as plans in TRL AMM agent - simple task, 1st come-1st serve pseudo-ADS-B=TCP/IP, test robustness of protocol w.r.t. communications failures test empirically with various scenarios arbitrary number of aircraft effects of timing, positions, speeds... test handoff from ATC, entry to SCA arrival/departure frequencies (Poisson distr.) ISIC-2003-50 Valasek, Ioerger, Painter COURSE OUTLINE SECTION I • Introduction To The Smart Cockpit And Free Flight • Free Flight Functionality • Architecture: Aircraft & Simulator, Software & Hardware SECTION II • Artificial Intelligence In The Cockpit • Intelligent Computing Techniques • Software Agents: The Key To Smart Cockpit Software SECTION III • Real Time Flight Simulation: The Basics • Real Time Flight Simulation For Free Flight SECTION IV • Selected Research Results • Conclusion: Summing It Up ISIC-2003-51 Valasek, Ioerger, Painter FLIGHT SIMULATION SYSTEM real-time flight simulator Fixed-base: Commander 700; AV-8A Harrier, F-5A Freedom Fighter SGI Onyx Reality II sim engine Networked bank of PC’s Center stick; sidestick 155o projected field of view 30 Hz refresh rate Programmable Head Up Display ISIC-2003-52 Valasek, Ioerger, Painter FLIGHT SIMULATION SYSTEM real-time flight simulator Head Down Displays (HDD) Reconfigurable CRT; touchscreen LCD Moving Map NAV Display Autopilot Glide slope capture Heading Altitude Pitch attitude FMS & Autopilot Interface Touch--Sensitive Screen Gear Handle Flight Management System (FMS) Jeppesen data base Pre-flight planning and enroute updating Moving map display ISIC-2003-53 Valasek, Ioerger, Painter HARDWARE ARCHITECTURE real-time flight simulator Bledsoe Video Signal Output Video Signal Output Gannon TCP/IP Simulation and External Display SGI Onyx Reality II Aikman UDP GAPATS & Agent System PC Favre SPiFI PC Warner ISIC-2003-54 Flutie Valasek, Painter, Ioerger SOFTWARE ARCHITECTURE Multiple PCs - Windows HEAD UP DISPLAY NAV FLIGHT INTERP SGI Machine - Unix HEAD DOWN DISPLAY-L EFS EXECUTIVE PILOT ADVISOR DATA OBJECT TCPIP GAPATS.CPP (GAPATS.H) ENVIRONMENT AGENT DATAOBJ AGENT EXECUTIVE SPIFI HEAD DOWN DISPLAY-R FLIGHT PLAN AGENT WEATHER AGENT TRAFFIC AGENT RADAR DATA (SIM) ADS-B DATA (SIM) ISIC-2003-55 Pilot Flight Controls FCS PROJECTED 3-SCREEN DISPLAY COMGEN (1) COMGEN (2) TRAG ROLCAS PITCAS YAWCAS EFS Dynamics Valasek, Ioerger, Painter SIMULATION SYSTEM SOFTWARE COMPONENTS Aircraft model • Six-degree-of-freedom dynamics model • CD & R model • FMS model (including Autopilot model) • Pilot model • ADS-B model • Additional research model ADS-B Flight Plan Aircraft Trajectory Flight Objectives Dynamics Model Pilot FMS(Autopilot)/Navigation CD & R Model Other Research Models Communication Controller ISIC-2003-56 Surveillance Valasek, Ioerger, Painter FLIGHT SIMULATION SYSTEM Combined GAPATS/Agents system functions like a simplified FMS Soft Pilot/FMS Interface (SPiFI) Pilot Commands SPiFI Agent System Autopilot Command Generator Engineering Flight Simulator GAPATS Tracker ISIC-2003-57 Valasek, Ioerger, Painter HEAD UP DISPLAY SYMBOLOGY current waypoint marker heading tape rate of climb airspeed altitude glide slope data text area runway outline pitch angle localizer data ISIC-2003-58 Valasek, Painter, Ioerger HEAD UP DISPLAY IN FLIGHT ISIC-2003-59 Valasek, Ioerger, Painter HEAD UP DISPLAY IN FLIGHT ISIC-2003-60 Valasek, Ioerger, Painter HEAD UP DISPLAY IN FLIGHT night and weather ISIC-2003-61 Valasek, Ioerger, Painter NAV/MAP DISPLAY SYMBOLOGY ISIC-2003-62 Valasek, Painter, Ioerger CONTROL EFFECTORS commercial air transport primary controls secondary controls outboard aileron flight spoilers leading edge flaps ground spoilers stabilator upper and lower rudders throttle inboard flap inboard aileron outboard flap Boeing 777-300 ISIC-2003-63 Valasek, Ioerger, Painter LIFT AND DRAG FORCES definitions Lift XB U1 Drag ZB Grumman F11F-1 Tiger ISIC-2003-64 Valasek, Ioerger, Painter AERODYNAMIC ANGLES definitions Sideslip Angle Angle-of-Attack VP XB XB VP XI Note: All Angles Shown Are Positive Grumman F11F-1 Tiger ISIC-2003-65 Valasek, Ioerger, Painter BODY AXIS component definitions Aerodynamic and Thrust Forces Acceleration of Gravity Aerodynamic and Thrust Moments Linear and Angular Velocities Note: Positive Signs Shown Reference 2-1 ISIC-2003-66 Valasek, Ioerger, Painter EULER ATTITUDE ANGLES definition Reference 2-2 ISIC-2003-67 yaw attitude angle pitch attitude angle roll attitude angle Valasek, Ioerger, Painter EQUATION SUMMARY Linear Motion m(U VR WQ ) mg x FA x FT x m(V UR WP ) mg y FA y FT y m(W UQ VP ) mg z FA z FT z Drag Equation " Drag Equation " " Sideforce Equation " Sideforce Equation " LiftLift Equation " Equation Angular Motion Assuming the x-z plane is plane of symmetry, i.e., I xy I yz 0 I xx P I xz R I xz PQ ( I zz I yy ) RQ L A LT Rolling Moment Equation I yy Q ( I xx I zz ) PR I xz ( P 2 R 2 ) M A M T Pitching Moment Equation I zz R I xz P ( I yy I xx ) PQ I xz QR N A N T ISIC-2003-68 Yawing Moment Equation Valasek, Ioerger, Painter FORCES AND MOMENTS shorthand notation Dividing by mass, each term becomes a longitudinal linear or angular acceleration. Letting moment 1 moment variable variable m , dimensional derivatives are the linear or angular acceleration per change in the associated motion variable. M is the pitch angular acceleration imparted to the airplane as the result of a unit change in angle-of-attack. 1 f X X u u X X X q q X E X F m 1 f Z Z u u Z Z Z q q Z E Z F m 1 m M u u M M M q q M E M F I YY E F E F E F There will be separate equations for the aerodynamic and thrust forces and moments. ISIC-2003-69 Valasek, Ioerger, Painter S & C DERIVATIVES wind tunnel testing OPTIONS Static testing Dynamic testing DIFFICULTIES Tests are not dedicated to estimation of performance Reynold’s number effects on drag dependent terms Cost of testing Wind tunnel availability ISIC-2003-70 Valasek, Ioerger, Painter STABILITY DERIVATIVES relative importance and prediction accuracy Relative Importance* Estimated Prediction Accuracy** CL α 10 ± 5% Cm α 10 10 CD α 5 10 4 40 7 40 CD α 1 50 CL u 8 20 Cm 8 20 6 CL q Cm Derivative CL α Cm α CD CD u u Cy β C l β Cn β Cy C l Cn Relative Importance* Estimated Prediction Accuracy** 7 ±20% 10 20 10 15 2 60 2 60 4 60 4 50 10 15 20 Cy p C lp Cn p 8 90 3 20 Cy 4 30 9 20 7 40 1 20 Cl p Cn p 9 25 q q Derivative p Reference 2-1 * 10 = Major, 5 = Minor, 0 = Negligible ** Using theoretical methods. With use of tunnel data, better accuracy can be achieved ISIC-2003-71 Valasek, Ioerger, Painter EQU ATION S OF M OTION observations The Equations of Motion are a set of 9 differential equations: First order Nonlinear Coupled Ordinary The Variables are: U, V, W, P, Q, R, , and The aerodynamic and thrust Forcing Functions: FAx , FT x , FA y , FT y , FAz , FTz , L A , LT , M A , M T , N A and N T are functions of: Velocity Angle-of-attack Sideslip angle Time Altitude Configuration ISIC-2003-72 Valasek, Ioerger, Painter AIRCRAFT DYNAMIC MODES standard longitudinal modes SHORT PERIOD The primary and most useful standard longitudinal dynamic mode. Second-order Stable, or unstable High frequency, well damped Exhibited mostly in angle-of-attack and body-axis pitch rate Specified in military flying qualities regulations Pitch maneuverability is based upon controlling and shaping this mode speed remains constant, angle-of-attack and pitch attitude vary Reference 3-1 ISIC-2003-73 Valasek, Ioerger, Painter AIRCRAFT DYNAMIC MODES standard longitudinal modes PHUGOID The secondary standard longitudinal dynamic mode (nuisance mode). Second-order Stable, or unstable Low frequency, very lightly damped Exhibited mostly in velocity and pitch attitude angle Specified in military flying qualities regulations Name derived from the Greek word for fly: “phugos” angle-of-attack remains constant, speed and pitch attitude vary Reference 3-1 ISIC-2003-74 Valasek, Ioerger, Painter AIRCRAFT DYNAMIC MODES example: Dutch roll mode Reference 3-1 ISIC-2003-75 Valasek, Ioerger, Painter AIRCRAFT DYNAMIC MODES example: roll mode Reference 3-1 ISIC-2003-76 Valasek, Ioerger, Painter AIRCRAFT DYNAMIC MODES example: spiral mode Reference 3-1 ISIC-2003-77 Valasek, Ioerger, Painter FLIGHT CONTROL PROBLEM statement The aircraft flight control design problem is to develop and implement an algorithm that closes the loop between the sensors and actuators, such that the aircraft accomplishes its mission, as dictated in the requirements and mission specification, with flying qualities deemed acceptable to the pilot. ISIC-2003-78 Valasek, Ioerger, Painter FLIGHT CONTROLLER DESIGN methodology REQUIREMENTS DEFINITION Based on intended mission and aircraft class military: specified entirely by Department of Defense; usually non-negotiable civilian: specified jointly between customer and manufacturer; negotiable Design iterations imply tradeoffs in requirements how do requirements translate into assumptions? MODELING “Goodness” origin validity assumptions accuracy suitability – dynamic order – state-space or frequency domain? Refinement ISIC-2003-79 Valasek, Ioerger, Painter CLASSES OF CONTROLLERS Autopilots Function: Provide pilot relief and special functions pilot commanded motion variables command shaping +- controller compensator Notes: controls vehicle motion vehicle sensors 1. Pilot inputs are outer-loop variables such as airspeed, heading, altitude, etc. 2. Limited-authority system. ISIC-2003-80 Valasek, Ioerger, Painter TYPICAL SPECIFICATIONS G.A. autopilot CONFIGURATIONS Clean: Climb: Power Approach: gear and flaps up gear up, flaps down gear down, flaps down AIRSPEED COMMAND AND HOLD SYSTEM Input type: hold commanded airspeed: maximum airspeed error: Maximum values: T 70% of maximum ramp anywhere in the range 0 h 10,000 feet 2 KIAS Rockwell Commander C700 ISIC-2003-81 Valasek, Ioerger, Painter TYPICAL SPECIFICATIONS G.A. autopilot PITCH ATTITUDE COMMAND AND HOLD Input type: Maximum positive commanded change in : Maximum negative commanded change in : 90% rise time on : Maximum overshoot on : Maximum values: e 5 degrees ramp 10 degrees at SLS altitude -8 degrees at SLS altitude 5 seconds 15% ALTITUDE COMMAND AND HOLD Input type: Commanded change in h: Maximum climb rate: Maximum error: Maximum values: e 5 degrees ramp anywhere in the range 0 h 15,000 feet Standard Instrument Climb of 500 feet/minute 50 feet of commanded altitude ISIC-2003-82 Valasek, Ioerger, Painter AUTOPILOTS pitch attitude command and hold Purpose: Feedback Variable(s): Typical Sensor(s): Primary Control Variable: Typical Compensators: ref Maintain commanded pitch attitude Pitch rate inner-loop and pitch attitude outer-loop Rate gyro and vertical gyro Elevator Gain; and second-order pole-zero canceling compensator ec e K ( ) e e e q q e c 1 S Kq vertical gyro ISIC-2003-83 Valasek, Ioerger, Painter AUTOPILOTS pitch attitude command and hold Response to commanded 5 degree ramp in pitch attitude angle Response to 5 deg theta command [deg] 6 4 2 0 0 2 4 6 8 10 0 2 4 6 8 10 0 2 4 Time (sec) 6 8 10 q [deg/sec] 4 2 0 -2 -4 e actual [deg] 4 2 0 -2 ISIC-2003-84 Valasek, Ioerger, Painter AUTOPILOTS altitude command and hold Feedback Variable(s): Typical Sensor(s): Primary Control Variable: Typical Compensators: H href Altitude, Pitch Attitude Rate Barometric Altimeter, Rate Gyro Elevator Gain, Lead-Lag, Proportional Derivative ea Kh eh lead-lag e e e elevator servo eRG aircraft dynamics h K PD ISIC-2003-85 Valasek, Ioerger, Painter AUTOPILOTS altitude command and hold PD = 10S+1 hdot (ft/s) 700 300 300 -100 -100 700 12 elevator (deg) h (ft) hcom (ft) 700 300 -100 0 25 50 75 time (sec) 100 ISIC-2003-86 0 -12 0 25 50 75 time (sec) 100 Valasek, Ioerger, Painter IMPLEMENTATION trim Case 1: Operation about a Single Trim Condition The following controller modes operate about a single trim condition: Yaw Damper Pitch Damper Roll Damper Wing Leveler Heading Hold Velocity Hold Operational Procedure: The pilot establishes the desired trim condition and engages the mode. If he wishes to change the trim condition, he disengages the mode, establishes the new trim condition, and re-engages the mode. ISIC-2003-87 Valasek, Ioerger, Painter IMPLEMENTATION trim Case 2 Automatic Transition between Two Trim Conditions Flight controllers that require an automatic transition between two trim conditions include: Velocity Command Rate of Climb Command Glide Slope Capture and Tracking Turn Rate Control Operating Procedures: The flight control system must be able to automatically handle trim changes as well as changes in the vehicle’s open-loop dynamics. ISIC-2003-88 Valasek, Ioerger, Painter COURSE OUTLINE SECTION I • Introduction To The Smart Cockpit And Free Flight • Free Flight Functionality • Architecture: Aircraft & Simulator, Software & Hardware SECTION II • Artificial Intelligence In The Cockpit • Intelligent Computing Techniques • Software Agents: The Key To Smart Cockpit Software SECTION III • Real Time Flight Simulation: The Basics • Real Time Flight Simulation For Free Flight SECTION IV • Selected Research Results • Conclusion: Summing It Up ISIC-2003-89 Valasek, Ioerger, Painter AGENT BASED HIERARCHICAL SYSTEM Executive Agent Weather Radar Data Ground Weather Service Other Traffic Info... Traffic Agent Weather Agent Other Weather Info. ... Flight ADS-B ATC Radar Plan Info. Overall Structure of Hierarchical Agent System ISIC-2003-90 Valasek, Ioerger, Painter WEATHER AGENT objective : safest and shortest route OTHER CONSTRAINTS Minimum segment length length of any segment in flight path cannot be less than this End point Maximum turning angle turns of angles greater than this are not allowed Minimum number of turns L Starting point METHOD USED L = segment length Modified A* Search Regions with intensity greater than 25 dBZ are set as inaccessible ISIC-2003-91 = turning angle Valasek, Ioerger, Painter TRAFFIC CD&R AGENT Inputs are ADS-B state vectors of aircraft in immediate airspace Rh ( x1 x2 ) 2 ( y1 y2 ) 2 Alert Zone Rv z1 z2 IF Rh <Rhp .AND. ABS(Rv)<Rvp THEN avoidance=.TRUE. ELSE avoidance =.FALSE. Detects potential traffic conflicts Non-Cooperative agent Protected Zone Only considers aircraft in alert zone Overlap of protected zones prohibited Size of zones determined by several spatial or temporal factors Calculates evasion maneuvers to avoid other protected zones Knowledge based expert system and optimal control ISIC-2003-92 Valasek, Ioerger, Painter EXECUTIVE AGENT Arbitrator between lower-level agents Intelligent behavior Fuzzy synthesis evaluation method Determines ultimate flight guidance submitted to: Weather Radar Image Weather Agent FMS Autopilot Pilot Flight Management System Rule-Based Arbitrator Weather Conflict Evaluation Traffic Conflict Evaluation TC Path Selection Agents Controller Traffic Situation ISIC-2003-93 Traffic Agent Valasek, Ioerger, Painter CONCLUSION So, You Really Think … That Software … Can Fly … An Airplane ? Yeah, Right !! ISIC-2003-94 Valasek, Ioerger, Painter FLIGHT SIMULATION LAB points of contact Director John Valasek, Ph.D. Aerospace Engineering Department Texas A&M University 3141 TAMU College Station, TX 77843-3141 Thomas R. Ioerger, Ph.D. Computer Science Department Texas A&M University 3112 TAMU College Station, TX 77843-3112 (979) 845-1685 valasek@aero.tamu.edu (979) 845-0161 ioerger@cs.tamu.edu FSL Web Page Web Page http://flutie.tamu.edu/~fsl http://faculty.cs.tamu.edu/ioerger/ ISIC-2003-95 Valasek, Ioerger, Painter