1. Integrated Design and Analysis Tools for Software-Based Control Systems Shankar Sastry (PI) Tom Henzinger Edward Lee University of California, Berkeley

2. 2 1. Model building and checking for hybrid systems 2. Embedded code generation from hybrid models 3. Multi-modal, hierarchical, and multi-vehicle control 4. Probabilistic hybrid systems and fault tolerance 5. Experimental rotorcraft platforms Research Thrusts

3. 3 1. From Hybrid Systems Models to Embedded Code 1a. Simulink to Giotto to E code 1b. Ptolemy to Embedded Java 2. Multi-vehicle Cooperative Control Focus of Presentation/Demos

4. 4 Model Requirements Platform Verification Implementation

5. 5 Model Requirements Platform Verification Implementation automatic (model checking) automatic (compilation)

6. 6 Model Requirements Platform Verification Implementation property preserving

7. 7 Component Requirements Platform Verification Implementation Component

8. 8 Requirements Platform Verification Implementation Composition Component no change

9. 9 A new paradigm to achieve Verifiability and Compositionality: The FLET (Fixed Logical Execution Time) Assumption Software Task read sensor input at time t write actuator output at time t+d, for fixed d

10. 10 Software Task read sensor input at time t write actuator output at time t+d, for fixed d d>0 is the task's "logical execution time" A new paradigm to achieve Verifiability and Compositionality: The FLET (Fixed Logical Execution Time) Assumption

11. 11 High-Confidence, Compositional Embedded Programming The control engineer specifies sampling rate d and permissible jitter j to solve the control problem at hand. The compiler ensures that d and j are met on a given platform (hardware resources and performance). If the compiler succeeds, then the code is time safe; otherwise the program is rejected. No "priority tweaking"!

12. 12 time ttime t+d possible physical execution on CPU buffer output A new paradigm to achieve Verifiability and Compositionality: The FLET (Fixed Logical Execution Time) Assumption

13. 13 output as soon as ready Contrast the FLET to Standard Practice

14. 14 -predictable timing and data behavior (no race conditions, minimal jitter) -portable, composable code (as long as the platform offers sufficient performance) Advantages of the FLET

15. 15 The E(mbedded) Machine: a virtual machine that executes tasks in real time under the FLET assumption. E (machine) code can be checked for time safetry. Giotto: a structured, high-level language for control applications which is compiled into E code. Implementations of the FLET UC Berkeley (Henzinger, Horowitz, Kirsch, Majumdar, Matic, Sanvido).

16. 16 UC Berkeley (Horowitz, Liebman, Ma, Koo, Sangiovanni-Vincentelli, Sastry). A Giotto-Based Flight Control System

17. 17 200 Hz 400 Hz 200 Hz 1 kHz A Giotto-Based Flight Control System

18. 18 1. Concurrent periodic tasks: -sensing -control law computation -actuating 2. Multiple modes of operation: -navigational modes (autopilot, manual, etc.) -maneuver modes (taxi, takeoff, cruise, etc.) -degraded modes (sensor, actuator, CPU failures) A Giotto-Based Flight Control System

19. 19 Mode 1 Mode 4Mode 3 Mode 2 Task S 400 Hz Task C 200 Hz Task A 1 kHz Task S 400 Hz Task C 200 Hz Task A’ 1 kHz Task C’ 100 Hz Task A 1 kHz Task S 400 Hz Task C 200 Hz Task A 2 kHz Task A” 1 kHz Condition 1.2 Condition 2.1 A Giotto-Based Flight Control System

20. 20 Host code e.g. C Glue code Giotto Functionality. -Reactivity. -Concurrency. Timing and interaction. -No time. -Sequential. A Giotto-Based Flight Control System

21. 21 The Giotto Tool Chain Simulink Model Giotto Program for task timing and interaction C Functions for tasks E CodePlatform Code Platform (minimal OS + hardware) E Machine invokes S/G Translator Giotto Compiler RTW Embedded Coder C Compiler S/G Simulator performance information guaranteed conformance (UC Berkeley, U Salzburg)

22. 22 Demo Tomorrow: The Giotto Development Kit The Giotto Development Kit 1.Giotto Compiler 2.Integrated Editor 3.E-code Viewer 4.E-code Simulator 5.Current work: -E-code analysis for time safety -E-code optimization UC Berkeley (Kirsch, Sanvido).

23. 23 Demo Tomorrow: Giotto-Based Embedded Control Examples An elevator controller: A controller for the Caltech vehicles:

24. Embedded Java Generation from Ptolemy Models Steve Neuendorffer Edward Lee Case Study: Caltech Vehicles

25. 25 Caltech Vehicles Wireless 802.11b Network Datagram with vehicle locations Controller RS-232 commands to fans

26. 26 A Hierarchical Heterogenous Model Measured physical parameters Discrete-event model convenient for events that do not occur at the same time

27. 27 A Hierarchical Heterogenous Model Data formatting Fan thrust map Continuous-time model good for physical hardware dynamics

28. 28 A Hierarchical Heterogenous Model Synchronous dataflow model convenient for signal processing and discrete-time aspects

29. 29 Stepwise Refinement of Simulation towards Implementation 802.11b RS-232

30. 30 Hardware-in-the-Loop 802.11b RS-232 Replace hardware-true simulation model with actual vehicle. Allows validation of hardware model aspects.

31. 31 Code Generation 802.11b RS-232 Replace controller simulation with embedded controller. Embedded Java Platform

32. 32 Directions Giotto code generation from Ptolemy Verify Giotto programs against hybrid automaton models Implement Softwalls algorithm on Caltech vehicles Dynamics similar to 2D aircraft dynamics, but safe for experimentation