Model-Based Virtual In-the-Loop-Test of Autonomous Systems: The FALTER Case


Published on

Presentation at the 2nd International Workshop on Model-driven Approaches for Simulation Engineering

(held within the SCS/IEEE Symposium on Theory of Modeling and Simulation part of SpringSim 2012)

Please see: for further details

  • Be the first to comment

  • Be the first to like this

No Downloads
Total views
On SlideShare
From Embeds
Number of Embeds
Embeds 0
No embeds

No notes for slide

Model-Based Virtual In-the-Loop-Test of Autonomous Systems: The FALTER Case

  1. 1. Model-­‐Based  So,ware  In-­‐the-­‐Loop-­‐Test  of   Autonomous  Systems The  FALTER  CaseAndreas  Bayha,  Franziska  Grüneis,  Bernhard  Schätzfor9ss  gGmbHMod4Sim@TMS/DEVS,  Orlando,  27.03.2012
  2. 2. FALTER  Project FALTER Mission Management Mission Data Execute Mission Result Information FALTER:  Flugeinheit  zur  Autonomen  Lage-­‐  und  Terrain-­‐Erkundung  Mission:  Autonomous  flight  for  in-­‐situ  indoor  analysis  (no  GPS  signal)  PlaBorm:  Quadrocopter  with  IF/US/IMU  Autonomy:  Online-­‐replanning  for  collision  avoidance2
  3. 3. FALTER:  HW-­‐PlaHorm Bluetooth Ultra Sonic WLAN I!C RC RoBoard I!C Compass Reciever USB PWM PWM Camera Safety switch RS232 PWM Gyros Battery FlightCrtl Accels Motors HW-­‐ConstrucJon:  Modular  PlaKorm  Sensors:  Incl.  gyroscope,  ultrasonic,  Pme-­‐of-­‐flight  camera,  alPmeter  Actuators:  Motors,  camera  CommunicaPon:  Mission  data/goal  informaPon,  emergency-­‐off  Flight  control:  COTS-­‐control  unit  for  quadrocopter    Mission  control:  Embedded-­‐qualified  GP  control  unit3
  4. 4. FALTER:  So,ware  VerificaJon Application Layer So,ware  Architecture Planing  HW-­‐abstracPon  layer Environment Data  ApplicaPon  layer:  Mission  funcPons Execution  Control:  Measuring  and  Control FALTER Data  ExecuPon:  Handling  of  flight  leg Control Command Sense  Planning:  (Re-­‐)Planning  of  mission  path   FA LT E R - H A L RoBoard-HAL FlighCtrl-HAL VerificaJon  Goals RoBoard FlightCrtl  Reliability:  Faults  of  plaBorms  Robustness:  SituaPons  in  environment Hardware & Abstraction Layer4
  5. 5. FALTER  Project:  IntegraJon  Test FALTER:  Complicated  and  risky  integraJon  test  Complex  state  space  (incl.  internal  model  of  environment)  Complex  environment  (incl.  plaBorm  faults,  unexpected  obstacles)  Safety  criPcal  funcPonality  (incl.  man  and  material)5
  6. 6. Virtual  IntegraJon:  Simulated  PlaKorm  and  Environment Virtual FALTER FALTER Planning Environment Model Execute FALTER Environment Environment Model Model Control Command & Sense Platform Model Platform Virtual  Commissioning:  Models  for  VerificaJon/ValidaJon  Pla$orm  model:  HAL,  hardware,  electronics&mechanics  of  system  FALTER:  Model  of  local  parameters  (e.g.,  posiCon,  speed)  Environment  model:  Physical  environment  of  system  FALTER:  Model  of  global  parameters  (e.g.,  walls,  obstacles)  Virtual  Commissioning:  ExecuCon  of  applicaCon  on  simulated  plaKorm  in   simulated  environment  6
  7. 7. UAV  SimulaJon:  State  of  the  Art  -­‐-­‐Tools UAV  SimulaJon:  Tools  for  Model  ConstrucJon 1. RC  Simulators:  SimulaCon  of  UAVs  for  RC  training  (e.g.,  FMS)  SiL-­‐Usage:  6-­‐DoF-­‐Models,  no  environment  and  sensor  models 2. Physics  simulators:  SimulaCon  mech./elec.  processes  (e.g.  SimScape)  SiL-­‐Usage:  Solids/fluids  models,  no  dyn.  environment  and  sensor  models7
  8. 8. SimulaJon:  Structure Structure: Modular Components Control   • Environment model: Walls, obstacles So]ware • Sensor model: Ultrasonic, time-of- flight, gyroscopes, accelerators PlaBorm • Actuator model: Flight mechanics, Model power electronics • Platform model: Preprocessing, flight Sensor Actuator control, API Model Model • Control software: Unmodified software Environment Model8
  9. 9. SimulaJon:  Actuator  model [] [] Walls Walls V e (m/s) V e (m/s) X e (m) Walls LED X (m) GAS e Position F (N) Body F (N) Body Gier xyz xyz DCM Euler Angles (rad) FC ACCEuler Angles Nick (rad) FC GYRO Roll DCMbe DCMbe Roboard V (m/s) V (m/s) b b Fixed (rad/s) Fixed (rad/s) Mass Mass M xyz (N m) M xyz (N m) d /dt d /dt A (m/s2 ) A (m/s2 ) b b 6DoF (Euler Angles) 6DoF (Euler Angles) GAS F YAW NICK ROLL M rad/s Flight Dynamics F Actuator  model:  Handling  of  flight  mechanics  Physics:  6DoF-­‐MoPon  model  Actuators:  TranslaPon  control  commands  via  power  electronics9
  10. 10. SimulaJon:  Environment  and  Sensor  Model z e1 e2 v yEnvironment  Model:  Walls,   Sensor  Modell:  Distance/PosiJon  Obstacles Measurement Surfaces:  One-­‐Vertex-­‐Dual-­‐Edges-­‐  Distance  Measurement:  PosiCon-­‐ Encoding  of  rectangles dependent  distance  list Walls,  Obstacles:  CombinaCon  of    PosiCon  detecCon:  Provision  of   surfaces 6DoF-­‐values    10
  11. 11. SimulaJon:  ImplementaJon Implementation: Matlab/Simulink • Simulation: Simulink Aerospace Toolbox, simulation components • Visualization: Simulink 3D Simulation (aka VR Toolbox) • Software inclusion: S-function via API of Platform Model11
  12. 12. SimulaJon:  Fault  Model EffecPve  Signal Signal+Dri] Signal+Dri]+Noise Fault Model: Support for generic classes of faults • Systemic faults, e.g., noise, drift • Sporadic faults, e.g., bit-flip, stuck-at • Parametrized faults, e.g., fail time, noise strength12
  13. 13. SimulaJon:  ApplicaJon Intended  Z-­‐Speed/ AlPtude EffecPve  Z-­‐Speed/ AlPtude Assumed  Speed/ AlPtude Application: In-the-Loop Test incl. Debugging by Simulation • Execution of software, simulation of platform and environment13
  14. 14. Test  Management  Test  Management  System:  Scenarios  Easily  reproducable  setups  for  in-­‐the-­‐loop  tests  Independent  combinaPon  of  noise,  dri],  blackout,  obstacles  ApplicaPon:  Reliability  (faults),  robustness  (obstacles)14
  15. 15. SimulaJon:  ApplicaJon15
  16. 16. Virtual  SiL  TestSimulaJon  Framework  for  UAVs Standard  architecture  (Environment,  sensors,  pla$orm,  actuators) Modular  components  (incl.  ultrasonic,  Cme-­‐of-­‐flight) Robustness/reliability  test  (incl.  obstacles,  sensor  defects,  Cming  faults) Debugging  support  (incl.  internal  environment  model)➡ Efficient  support  of  early  and  low-­‐risk  validaCon/verificaCon➡ LimitaCon  due  to  degree  of  details  (e.g.,  energy  effects,  surface  properCes) More  InformaJon: at our site at www.fortiss.org16