s v v
play

S V V .lu software verification & validation Automated - PowerPoint PPT Presentation

S V V .lu software verification & validation Automated Testing of Autonomous Driving Assistance Systems Lionel Briand VVIoT, Sweden, 2018 Collaborative Research @ SnT Research in context Addresses actual needs Well-defined


  1. S V V .lu software verification & validation Automated Testing of Autonomous Driving Assistance Systems Lionel Briand VVIoT, Sweden, 2018

  2. Collaborative Research @ SnT • Research in context • Addresses actual needs • Well-defined problem • Long-term collaborations • Our lab is the industry 2

  3. Software Verification and Validation @ SnT Centre • Group established in 2012 • Focus: Automated, novel, cost- effective V&V solutions • ERC Advanced Grant • ~ 25 staff members • Industry and public partnerships 3

  4. Introduction 4

  5. Autonomous Systems • May be embodied in a device (e.g., robot) or reside entirely in the cyber world (e.g., financial decisions) • Gaining, encoding, and appropriately using knowledge is a bottleneck for developing intelligent autonomous systems • Machine learning, e.g., deep learning, is often an essential component 5

  6. Motivations • Dangerous tasks • Tedious, repetitive tasks • Significant improvements in safety • Significant reduction in cost, energy, and resources • Significant optimization of benefits 6

  7. Autonomous CPS • Read sensors, i.e., collect data about their environment • Make predictions about their environment • Make (optimal) decisions about how to behave to achieve some objective(s) based on predictions • Send commands to actuators according to decisions • Often mission or safety critical 7

  8. A General and Fundamental Shift • Increasingly so, it is easier to learn behavior from data using machine learning, rather than specify and code • Deep learning, reinforcement learning … • Assumption: data captures desirable behavior, in a comprehensive manner • Example: Neural networks (deep learning) • Millions of weights learned • No explicit code, no specifications • Verification, testing? 8

  9. Many Domains • CPS (e.g., robotics) • Visual recognition • Finance, insurance • Speech recognition • Speech synthesis • Machine translation • Games • Learning to produce art 9

  10. Testing Implications • Test oracles? No explicit, expected test behavior • Test completeness? No source code, no specification 10

  11. CPS Development Process Software-in-the-Loop Hardware-in-the-Loop Model-in-the-Loop Stage Stage Stage Architecture modelling Functional modeling: Deployed executables on • Structure • Controllers target platform • Behavior • Plant • Traceability • Decision System engineering modeling Hardware (Sensors ...) (SysML) Analog simulators Continuous and discrete Analysis: Testing (expensive) Simulink models • Model execution and testing Model simulation and • Model-based testing testing • Traceability and change impact analysis • ... (partial) Code generation 11

  12. MiL Components Plant Sensor Decision Actuator Controller 12

  13. Opportunities and Challenges • Early functional models (MiL) offer opportunities for early functional verification and testing • But a challenge for constraint solvers and model checkers: • Continuous mathematical models, e.g., differential equations • Discrete software models for code generation, but with complex operations • Library functions in binary code 13

  14. Automotive Environment • Highly varied environments, e.g., road topology, weather, building and pedestrians … • Huge number of possible scenarios, e.g., determined by trajectories of pedestrians and cars • ADAS play an increasingly critical role • A challenge for testing 14

  15. Testing Advanced Driver Assistance Systems 15

  16. Objective • Testing ADAS • Identify and characterize most critical/risky scenarios • Test oracle: Safety properties • Need scalable test strategy due to large input space 16

  17. Automated Emergency Braking System (AEB) Decision making Vision Brake (Camera) Controller Objects’ position/speed “Brake-request” Sensor when braking is needed to avoid collisions 17 17

  18. Example Critical Situation • “AEB properly detects a pedestrian in front of the car with a high degree of certainty and applies braking, but an accident still happens where the car hits the pedestrian with a relatively high speed” 18

  19. Testing via Physics-based Simulation 19

  20. Simulation Simulator Ego Vehicule Inputs Environment (physical plant) - the initial state of the physical plant and the Dynamic mobile objects mobile environment models objects Pedestrians - the static environment aspects Other actuators Vehicules Outputs static aspects sensors Time-stamped vectors for: - the SUT outputs - Road SUT - the states of the physical - Traffic sign plant and the mobile cameras - Weather environment objects Feedback loop 20

  21. Our Goal • Developing an automated testing technique for ADAS • To help engineers efficiently and effectively explore the complex test input space of ADAS • To identify critical (failure-revealing) test scenarios • Characterization of input conditions that lead to most critical situations 21

  22. ADAS Testing Challenges • Test input space is large, complex and multidimensional • Explaining failures and fault localization are difficult • Execution of physics-based simulation models is computationally expensive 22

  23. Our Approach • Effectively combine evolutionary computing algorithms and decision tree classification models • Evolutionary computing is used to search the input space for safety violations • We use decision tress to guide the search-based generation of tests faster towards the most critical regions, and characterize failures • In turn, we use search algorithms to refine classification models to better characterize critical regions of the ADAS input space 23

  24. AEB Domain Model Snow Normal « enumeration » « enumeration » WeatherC - snowType: CurvedRadius (CR) VisibilityRange {{OCL} self.fog=false SnowType Rain - 5 - 10 - 15 - 20 - 10 - 20 - 30 - 40 - 50 implies self.visibility = “300” - 25 - 30 - 35 - 40 - 60 - 70 - 80 - 90 - 100 - rainType: and self.fogColor=None} Weather - 110 - 120 - 130 - 140 RainType « enumeration » - 150 - 160 - 170 - 180 - visibility: Curved RampHeight (RH) - 190 - 200 - 210 - 220 VisibilityRange - radius: - 4 - 6 - 8 - 10 - 12 - 230 - 240 - 250 - 260 - fog: Boolean 1 CurvedRadius - 270 - 280 - 290 - 300 - fogColor: « enumeration » FogColor « enumeration » Test Ramped SnowType Scenario - height: RainType - ModerateSnow 1 Road RampHeight - simulationTime: - ModerateRain - HeavySnow - frictionCoeff: Real - HeavyRain Straight - VeryHeavySnow Real - timeStep: Real - VeryHeavyRain - ExtremeSnow 1 - ExtremeRain « enumeration » 1 Vehicle 1 FogColor Pedestrian AEB Output 1 v c 1 1 - v0 : Real - DimGray - : Real x p 0 0 - : TTC: Real v 1 Position - Gray y p - : Real - : certaintyOfDetection: v 2 0 - DarkGray - x: Real Real - : Real v p - Silver - y: Real 0 - : braking: Boolean v 3 θ p - LightGray - :Real 0 1 - None 1 Position vector Static input Output functions 1 1 F 1 - : Real Dynamic input Mobile F 2 - : Real object Output

  25. Search-Based Software Testing • Express test generation problem portion of input domain as a search problem denoting required test data • Search for test input data with certain properties, i.e., randomly-generated inputs constraints Input domain • Non-linearity of software (if, Random search may fail to fulfil low-probability loops, …): complex, discontinuous, non-linear search spaces (Baresel) Genetic Algorithm Fitness • Many search algorithms (metaheuristics), from local search to global search, e.g., Hill Climbing, Simulated Annealing and Genetic Input domain Algorithms Genetic Algorithms are global searches, sampling man 25 “Search-Based Software Testing: Past, Present and Future” Phil McMinn

  26. Multiple Objectives: Pareto Front F 1 Pareto front x Individual A Pareto dominates individual B if A is at least as good as B Dominated by x in every objective and better than B in at least one objective. F 2 A multi-objective optimization algorithm (e.g., NSGA II) must: • Guide the search towards the global Pareto-Optimal front. • Maintain solution diversity in the Pareto-Optimal front. • 26

  27. Decision Trees All points Count 1200 “non-critical” 79% “critical” 21% RoadTopology RoadTopology ( CR = 5 , ( CR = [ 10 − 40 ]( m )) Straight , RH = [ 4 − 12 ]( m )) Count 564 Count 636 “non-critical” 59% “non-critical” 98% “critical” 41% “critical” 2% θ p θ p 0 < 218 . 6 � 0 > = 218 . 6 � Count 412 Count 152 “non-critical” 49% “non-critical” 84% “critical” 51% “critical” 16% v p v p 0 < 7 . 2km / h 0 > = 7 . 2km / h Count 230 Count 182 “non-critical” 31% “non-critical” 72% “critical” 69% “critical” 28% Partition the input space into homogeneous regions 27

Recommend


More recommend