Automotive CPS Integration and Experimental Platform

AUTOMOTIVE CPS: INTEGRATION CHALLENGES AND EXPERIMENTAL PLATFORM In Support of NSF CPS Science of Integration Project ECS Process, Methods and Tools Group Electrical and Controls Integration Lab General Motors Global R&D

2011 GM Support & Interactions GM provided $50K funding Site Visits o GM hosted two visits o GM visits to Notre Dame and Vanderbilt GM provided 2 summer internships o Emeka Eyisi, Vanderbilt Univ. o Dimitris Spyropoulos, Univ. of Maryland 2/27/2012 NSF Site Review at Vanderbilt University 2

Vehicle Control System Development ACC CTD EngCyc LDW KLE IMC LXC VRP DFI model translation /composition tool Integration ACC-v1 ACC-v1 ACC-v1 ACC-v1 ACC-v1 CTD-v1 ACC-v1 ACC-v1 EngC-v1 Natural language Use-case diagram Block diagram, etc. Word Excel DOORS Requirements ACC-v1 ACC-v1 LDW-v1 ACC-v1 ACC-v1 KLEC-v1 ACC-v1 ACC-v1 IMC-v1 Diff. Equation,, State- based formalism, SDF, etc. Math emati ca State- mate ACC-v1 ACC-v1 LXC-v1 ACC-v1 ACC-v1 VRP-v1 Analysis ACC-v1 ACC-v1 DFI-v1 development process model transformation Continuous + Discrete tool integration Matlab SL/SF Simulation UML, UML-Profile, ADL (AADL, EAST- EEA), AutoSAR, etc Ascet SD Rhapsody Design dSpace Target- Link ACC-v1 ACC-v3 KLEC-v2 KLEC-v5 LXC-v3 LXC-v3 VRP-v2 VRP-v1 C/C++ Implementation (RP vs. Target) Code Libraries Programming Env. (Tornado, MS Visual Studio, etc) EngC-v1 EngC-v1 DFI-v5 DFI-v3 CTD-v1 Chevrolet Buick 2/27/2012 NSF Site Review at Vanderbilt University 3

Automotive Evaluation Platform Integration Scenario 1 o Product configuration plug-play features Integration Scenario 2 o Impact of new E/E technology plug-play platform Integration Scenario 3 o New property plug-play property Integration Scenario 4 o Runtime reconfiguration based on modes dynamic plug-play 2/27/2012 NSF Site Review at Vanderbilt University 4

Progress Summary Virtual experimental platform construction Physical-control-software-platform impact investigation Engineering Process and Tool Integration o System modeling and process modeling Collaboration and Support o Telecons, internships, cash and in-kind contributions, exchange visits 2/27/2012 NSF Site Review at Vanderbilt University 5

Experimental Platform TrueTime ACC Componenti- zation Fredo Fredo Fredo TTE Switch Virtual Ethernet Fredo Fredo 2/27/2012 NSF Site Review at Vanderbilt University 6

Current Design Method Stakeholder Requirements User Acceptance Test Bill of Features System Requirements System Integration Test Hardware Subsystem Requirements Software Subsystem Requirements Subsystem Integration Software Test Subsystem Integration Hardware Test Hardware Subsystem Design Software Subsystem Design Software Subsystem Wht Box Test Hardware Subsystem Internal Test Hardware Component Implementation Software Component Implementation Software Unit Test Hardware Component Test ECU Allocation Vehicle Layout Control development o Algorithm development o Discretization Integration o Allocation and scheduling Software development o Component creation o Software architecture 2/27/2012 NSF Site Review at Vanderbilt University 7

Research Challenges To-Be Addressed Modeling relations between cyber and physical system o What parameters of cyber affect what properties of physical system o Simulations help get insight of how different strategy impact on physical systems Quantifying relations between control and software start from simple case o Control architecture: hierarchical control o Software architecture: time-triggered/periodic Mappings between control system and software system o Control partition matters o Software packaging matters o Software architecture matters o Runtime parameters (schedule, period, priority, etc.) matters 2/27/2012 NSF Site Review at Vanderbilt University 8

Observation: Physical Systems - Control Same ABS control algo Different vehicle types Different load (ext. 600 lb) 2/27/2012 NSF Site Review at Vanderbilt University 9

Observations: Software Architecture - Control Impact of different implementation strategies temporal firewall, temporal interfaces, hierarchical scheduling, dynamical feedback scheduling, etc Per = 10, wc completion time = 1~5 Per = 6, wc completion time = 1~5 2/27/2012 NSF Site Review at Vanderbilt University 10

Methods Supporting Integration with Variance Control property sensitive to physical system variance o Control parameters allow to adapt to physical system change Preserving control with SW variance o Correlations between sampling period and delay o Control-driven software architecture selection o Interferences of control integration at software/platform level Relations (mathematical if possible) among physical config., control arch., and software arch. o Hierarchical control o Integrated software architecture and hierarchical scheduling 2/27/2012 NSF Site Review at Vanderbilt University 11

System Engineering Process Assess SysML as modeling language for system engineering o Existing process modeling and enforcement o Process extension safety process 2/27/2012 NSF Site Review at Vanderbilt University 12

Tool Integration Tool integration for simulation o CarSim + Simulink/Stateflow + TrueTime o Applied to ACC (ACC, FSR-ACC, LXC-ACC, SLC-ACC) Tool integration for design o Intended to support control-software co-simulation o Simulink + Rhapsody: failed due to tool interfacing 2/27/2012 NSF Site Review at Vanderbilt University 13

Collaborations and Support GM committed support o Summer interns: Emeka Eyisi (VU), Dimitris Spyropoulos (Maryland) o GM cash contribution: $50K for 2011 Visits and meetings o March, Nicholas visit: experimental platform o June, Vanderbilt: experimental platform o July, Notre Dame: vehicle control system o August, Panos team at GM: control features for experiment 2/27/2012 NSF Site Review at Vanderbilt University 14

THANK YOU 15 2/27/2012 NSF Site Review at Vanderbilt University