Cooperative Control Challenges for Aerial Vehicles

Transcription

1 Cooperative Control Challenges for Aerial Vehicles Gidance Reference + Model - Adaptive Control Baseline + + Airframe Atopilot Egene Lavretsky, Ph.D. Senior Technical Fellow Boeing Research & Technology The Boeing Company AFOSR Workshop on Adversarial and Stochastic Elements in Atonomos Systems Washington, DC, March 23 24,

2 Presentation Overview Introdction Single vehicle Mltiple Vehicles Waves of mltiple vehicles Control Challenge Problems Single Vehicle in Close-Copled Flight with a Leader Simplified Flight Dynamics Applications Atonomos Aerial Refeling, (AAR) Atonomos Formation Flight, (AFF) Mltiple Unmanned Aerial Vehicles, (UAVs) Simplified UAV dynamics Dbins' car Dbins' aircraft Collaborative Control in Hostile / Adversarial Environment Task allocation Path planning Cooperative attack Intelligence Srveillance and Reconnaissance, (ISR) Ultimate Challenge Waves of mltiple UAVs prosecting mltiple targets in ncertain environment Conclsions 2

3 Close-Copled Flight Dynamics and Control Aerodynamic Phenomenon Unknown nsteady flow field behind lead aircraft Lead aircraft wingtip vortices inflence trailing aircraft aerodynamic forces and moments Longitdinal separation changes are niform within 5 aircraft body length Vertical and Lateral Separation indced aerodynamic drag rolling moment Relative bank angle Drag Redction (φ = 0) 0.03 Indced Rolling Moment (φ = 0) C D form / C D ΔY y Δ Z ΔC l Δ Z ΔYy 3

4 Close-Copled Flight: System Dynamics Trailing aircraft in close-copled formation (relative dynamics) 2 Control Inpts Thrst for Airspeed Differential Aileron for Roll Unknown Constant Parameters Throttle effectiveness Roll damping Aileron effectiveness d = V airspeed V = T δ D y δ T y = gn ϕ = p p z T ϕ a (, ϕ) bank angle roll rate Diff Aileron indced rolling moment p = L p+ L δ +ΔL y δ Thrst a indced drag force (, ϕ ) Unknown fnctions Vortex indced aerodynamics Drag force D( y, ϕ ) Rolling moment ΔL( y, ϕ ) Applications Atonomos Formation Flight Atonomos Aerial Refeling C D form / C D Atonomos Aerial Refeling, (AAR) (, ϕ) = (, ϕ) (, ϕ ) (, ϕ ) D y D y ΔL y = ΔL y Drag Redction (φ = 0) ΔY y Δ Z Atonomos Formation Flight, (AFF) ΔC l Indced Rolling Moment (φ = 0) ΔY y Δ Z

5 Close-Copled Flight: AFF Control Challenge Benefit Flying in the vortex field redces indced drag by 20 25% range extension less fel Control Challenges wingtip vortex indced ncertainties nknown / nsteady vortex location Control Goals Bonded tracking Vortex seeking drag redction nknown vortex field F/A-18 Formation Flight NASA Dryden, 2000 d y longitdinal separation lateral separation 5

6 UAV Dynamics for Control Design Dbins' Car V = ( x, y) T x = V cosψ y = Vsinψ ψ = ψmax ψ V = V max V trn control max trn rate speed control max thrst y x ψ 2 control inpts Speed control, (thrst) limited between 0 and 1 Heading control, (rate of trning) limited between 0 and 1 2 controlled otpts (x, y) positions 6

7 UAV Dynamics for Control Design (contined) Dbins' Aircraft x = V cosγ cosψ y = V cosγ sinψ z = Vsinγ γ = γmax γ ψ = ψmax ψ V = V max V climb control max climb rate trn control max trn rate speed control x z ψ γ y V = ( x, y, z ) T 3 control inpts Speed control, (thrst) max thrst limited between 0 and max thrst available Heading and flight path controls, (trn and climb rates) limited between 0 and 1 3 controlled otpts (x, y, z) positions 7

8 Control Challenge: Single UAV Path Planning Single vehicle operating in ncertain environment Finding shortest path in finite graph System dynamics Simple model: Dbins' car / aircraft model Other Optimality criterion not a distance-like measre doesn t satisfy the trianglar ineqality instant connection cost matrix is not-symmetric Optimal path / rote = k+ 1 node precedence constraints obstacles and pop-p threats avoidance collision avoidance with other vehicles x k x = V cosψ y = Vsinψ ψ = ψ max ψ x = V cosγ cosψ y = V cosγ sinψ z = Vsinγ V = V max γ = γ V max ψ = ψ max V = V max γ ψ V Constrained Shortest Path Problem 8

9 Control Challenge: Task Allocation Dynamic Optimization Problem M vehicles N Tasks, (M < N) nmber of assets (M) and tasks (N) may change UAVs Tasks Goal maximize total assignment benefit accont for vehicle capabilities 1 2 Online soltion comptation i M Constrained Optimal Assignment Problem 9

10 Control Challenge: Cooperative Attack Mltiple UAVs Prosecting Mltiple Targets Task allocation Constraints individal vehicle capability time-critical targets task precedence relative timing Path / Rote planning Obstacle & Collision Avoidance Sensing, Estimation and Information Sharing Uncertain & hostile environment Limited commnication data links Constrained Mlti-Vehicle Shortest Path and Travelling Salesman Problems 10

11 Ultimate Challenge Single UAV Mltiple UAVs Waves of Mltiple UAVs Performing tasks to accomplish higher level objectives Collaborative Control Tasks in Uncertain Environment Cooperative Strike, Intelligence Srveillance Reconnaissance (ISR) Task allocation, Rote planning, and Obstacle Avoidance Task precedence and timing constraints Prosection of time critical targets Exection in the presence of ncertainty cased by adversarial actions threat location, anti-tactics degraded commnications Information Management Between Platforms Spport data fsion / estimation and collaboration limited available data links interrptions of service degraded performance 11

12 Conclsions Control Challenge Problems for Aerial Vehicles single UAV mltiple UAVs waves of mltiple UAVs Need Close-to-Real-Time Control Soltions Freqently generate optimal task assignment and mlti-vehicle rotes in the presence of: ncertain and hostile environment battle damage limited commnications Assign and prosecte time-critical targets Plan Find Fix Track Target Engage Shorten the Kill Chain Assess Commnicate 12

13 13