Software Engineering of Robots

Transcription

1 Software Engineering of Robots Ana Cavalcanti Jon Timmis, Jim Woodcock Wei Li, Alvaro Miyazawa, Pedro Ribeiro University of York December 2015

2 Overview One of UK eight great technologies: robotics and autonomous systems. 13 billion global market predicted for 2025 Safety: numerous applications of concern Autonomous vehicles Home automation Full verification is beyond the of the art Among other concerns: verification of controller software Models enable validation of requirements

3 Current approach to development 1st phase: Abstract model 2nd phase: Simulation controller hardware discrete 3rd phase: Implementation low-level + robot +

4 What do we want to do? 1st phase: Abstract model 2nd phase: Simulation controller hardware discrete 3rd phase: Implementation low-level + robot +

5 What do we want to do? 1st phase: Abstract model 2nd phase: Simulation controller hardware discrete 3rd phase: Implementation low-level + robot +

6 What do we want to do? 1st phase: Abstract model 2nd phase: Simulation controller hardware discrete 3rd phase: Implementation low-level + robot +

7 What do we want to do? 1st phase: Abstract model hardware + model controller hardware discrete 2nd phase: Simulation 3rd phase: Implementation low-level + robot +

8 What do we want to do? 1st phase: Abstract model hardware + model controller hardware discrete 2nd phase: Simulation 3rd phase: Implementation low-level + robot +

9 What do we want to do? 1st phase: Abstract model hardware + model controller hardware discrete 2nd phase: Simulation 3rd phase: Implementation low-level + robot +

10 How do we want to do it? like in UML (subset)

11 How do we want to do it? like in UML-MARTE like in UPPAAL

12 How do we want to do it? like in PRISM

13 How do we want to do it?

14 How do we want to do it?

15 How do we want to do it? hybrid like in SpaceEx hybrid hybrid hybrid

16 How do we want to do it? hybrid hybrid hybrid hybrid

17 And now to s and programs generation verified library controller annotated with time restrictions

18 And now to s and programs generation model transformation verified library controller annotated with time restrictions

19 Traceability is very important. automatic identification of affected parts/ change propagation controller automatic identification of affected parts/ change propagation controller deployment change OhCircus Time

20 And now to s and programs hybrid generation model transformation ( traceable) verified library controller annotated with time restrictions robot

21 In more detail Textual Editor RoboTool Notation Timed Notation RoboTool RoboSim UPPAAL, Theorem Proving etc Timed Semantics Semantics FDR, Theorem Proving etc Simulation Analysis Analysis

22 Plan of work Five themes Core Time Probability: refinement is very challenging Environment: no need for new abstractions generation is very challenging abstractions Integration theory of integration library major case study: boats and drones

23 Plan of work Five themes Core Time Probability: refinement is very challenging Environment: no need for new abstractions generation is very challenging abstractions Integration theory of integration library major case study: boats and drones

24 Conclusions A lot to do Theory: UTP Practice: new languages (formal, diagrammatic, API), their implementations Verification: compositional, scalable, traceable Our distinctive vision Notations akin to those already used Sound integration Full life cycle The theory is that of cyber-physical systems.