Welcome and Introduction

Jon Friedman, The MathWorks         

Jon Friedman will kick off the conference with a discussion of the state of embedded systems development in the automotive industry. His talk will highlight how automotive engineers can use Model-Based Design to meet the conflicting goals of developing more complex systems in a shorter time period with fewer prototypes.

Jon Friedman leads the MathWorks automotive marketing effort to support industry adoption of Model-Based Design and MathWorks tools. Before joining The MathWorks, Jon held various positions at Ford Motor Company that included embedded systems researcher at the Ford Research Lab, product launch leader at plants across North America, and electrical engineering system and integration supervisor. While at Ford, Jon work with a team of researchers to develop and document tools for computer-aided control system design. In addition to his experience with Ford, Jon worked as an independent consultant on projects for Delphi, General Motors, Chrysler and the U.S. Tank-automotive and Armaments Command. Jon holds a B.S.E., M.S.E. and Ph.D. in Aerospace Engineering, as well as a M.B.A, all from the University of Michigan.

Keynote: Meeting the Challenges After the Perfect Storm

Dr. David Cole, Chairman of the Center for Automotive Research

At the 2006 CAR Executive Briefing, David Cole described the perfect storm facing the automobile industry — rising costs, increased competition, erosion of pricing power, and reduced time to market.  In this keynote, David will discuss how companies are adopting new engineering processes and advanced tools and technologies to increase their productivity and reduce their costs. This change is causing a transformational shift from the physical world to the virtual world.  The current generation of engineers develops their designs in the 1's and 0's of modern computer aided design software. An increasing amount of intellectual property stays in the 1'’s and 0's as software algorithms that control everything from modern emissions safety standards to the position of the driver's seat.  And just as the automotive industry previously replaced steel and iron prototypes with virtual models, it is now moving electronics and software development to a virtual environment. To address this need, engineers are increasingly relying on software to create the "look and feel" of the vehicle's DNA. All of this is done to provide systems and software to differentiate the customer experience, which now carries the same strategic importance as the interior and exterior design.

Dr. David E. Cole is the Chairman of the Center for Automotive Research (CAR) in Ann Arbor, Michigan. He was formerly Director of the Office for the Study of Automotive Transportation (OSAT) at the University of Michigan Transportation Research Institute. He has worked extensively on internal combustion engines, vehicle design, and overall automotive industry trends. Dr. Cole received his B.S.M.E., M.S.M.E., and Ph.D. from the University of Michigan.

Dr. Cole's recent research has focused on strategic issues related to the restructuring of the North American industry and trends in globalization, technology, market factors, and human resource requirements.

Best Practices for Establishing a Model-Based Design Culture

Paul Smith, The MathWorks

The transition to Model-Based Design requires careful management, both to demonstrate its short-term benefits and establish a culture that enables the full realization of the theoretical benefits of this approach. In this session, we will introduce the concepts of Model-Based Design, highlight some of its benefits, and discuss in detail 10 best practices for adopting Model-Based Design across an organization. These best practices have been gleaned from successful and not-so-successful transformations to Model-Based Design at companies from a variety of industries.

Paul Smith is the Senior Manager of North American Automotive Consulting at The MathWorks. He is responsible for managing the Novi, MI office as well as North American-based consulting businesses. Paul is the Challenge X technical advisor for The MathWorks, a three-year student competition sponsored by General Motors and the U.S. Department of Energy, heading up the mentoring and training programs. He has over 20 years of experience working in the automotive engineering industry; his experience at Ford Motor Company includes working on production powertrain control systems development, diesel engine controls, and automotive mechanical design and manufacturing processes. Paul holds an M.S.E.C.C.S from Wayne State University and a B.S.E.E. from Michigan Technological University.

Use of MathWorks Tool Chain for Proof-of-Concept Projects

Lev Vitkin, Delphi

Established automotive engineering teams with well integrated tool chains develop rapid prototyping and production control projects on detailed schedules with a defined budget. A different type of project, often called a proof-of-concept project, seeks to gauge market interest for new ideas and technologies. Many proof-of-concept projects are developed by a hastily put together, short-lived team of engineers with various skill sets. These projects have severe limitations on development time and often require the incorporation of different tools and technologies into one package. The selection of the tools that not only support the development of individual projects’ components, but also communicate with other tools from the same project can significantly minimize project development time.

The tool chain from The MathWorks has attributes that make it attractive for the development of proof-of-concept projects for automotive applications. This paper describes the use of MATLAB, Simulink, and Stateflow, along with Data Acquisition Toolbox and Real-Time Workshop Embedded Coder for the implementation of the new human-machine interface concept, including a vehicle with Integrated Center Module with dual-zone control and display.

Challenges of Creating Complex Integrated Vehicle System Models

Poyu Tsou, Ford Motor Company

Model-based system engineering (MBSE) is critical to the development of advanced vehicle systems.  Efficiently creating full vehicle system models for an MBSE process requires integration of models from different sources within the corporate modeling community. Such integration is a challenging task for the vehicle system model developers. This paper presents examples that show key aspects of creating full vehicle system models through integration of models from various sources and modeling environments. The MATLAB and Simulink product families are used for overall vehicle model integration. In the process described, component and subsystem models from various sources, including Dymola, GT-Power, and in-house proprietary simulation tools, are incorporated into full vehicle models to provide high fidelity simulation capability for development of hybrid electric vehicle systems. The challenges and problems encountered while creating such models are discussed along with solutions and direction for future work.

Modeling a Hybrid Electric Vehicle and Controller to Optimize System Performance

Steve Miller, The MathWorks

There is increasing pressure on the automotive industry to create vehicles that take advantage of alternative energy sources. The ever increasing cost of fuel and the more stringent emission standards require that new technologies be developed to meet those needs.  At the same time, the automotive industry needs to keep the consumer happy by continuing to meet its established performance requirements, while continuing to benefit from the infrastructure (such as gas stations and fuel delivery) already in place.  Such a system requires that the analysis of the design take place at the system level in order to understand the effects of integrating multidomain systems. In this talk, we address these issues with a discussion of how Model-Based Design was used to model and develop the electrical system of a hybrid-electric vehicle.

Steve Miller joined the Design Automation Marketing group at The MathWorks in 2005. Prior to that, Steve worked at Delphi Automotive in Braking Control Systems and at MSC.Software Adams consulting in various capacities at Ford, GM, Hyundai, BMW, and Audi. Steve has a B.S. in Mechanical Engineering from Cornell University and an M.S. in Mechanical Engineering from Stanford University.

JMAAB Vehicle Model Architecture and Two-Way Connection

Hisahiro Ito, Toyota Motor Corporation

JMAAB Physical Modeling Working Group recently published a Simulink style guide, Guidelines for Model Descriptions (Plant Models). An important part of the guide deals with vehicle model architecture: top-down, or how system models are broken down into components; and bottom-up, or how component models are joined together to build up systems. In the JMAAB guidelines, physical component models are connected using physical connection lines (as used in MathWorks physical modeling tools) rather than Simulink signal lines, in order to make the models as simple as possible to assemble, understand, and maintain. Component models developed using the traditional signal flow block diagram representation can also be inserted into the architecture, thanks to the new Two-Way Connection block, which bundles input and output signals into a physical connection line. In this paper, we present the JMAAB Vehicle Model Architecture and discuss the rationale and benefits of the two-way connection.

Automated Report Generation Solution for Analysis of Worldwide Fuel Cell Vehicle Fleet

Taylor Roche, Tim McGuire, and Andreas Weinberger
DaimlerChrysler, REDNA

DaimlerChrysler is an industry leader in the development and deployment of fuel cell vehicles. With more than 100 vehicles being driven worldwide at locations that include the U.S., Singapore, Japan, Germany, China, and Australia, DaimlerChrysler currently operates the world’s largest fuel cell vehicle fleet. Each vehicle is equipped with a powerful telematics system that records fuel cell specific vehicle operation data. This onboard data acquisition system uploads vehicle performance data to a central database via a DaimlerChrysler wireless infrastructure, where it is stored for further analysis.

The central database is a key component to the continued development of both fuel cell vehicle technology and a hydrogen infrastructure. DaimlerChrysler engineers are utilizing MathWorks software to analyze vehicle performance and guide decisions surrounding the development of fuel cell technology. DaimlerChrysler engineers have translated nearly two years of data into unique knowledge of hydrogen infrastructure needs, system diagnostics, and overall vehicle performance through statistical analyses and automated report generation.

The ability to condense vast amounts of data into obvious and insightful reports has become a valuable tool for all levels of management and is, therefore, shaping the future of fuel cell and hydrogen technologies.

Architecting Embedded Software Using Model-Based Design

Alan Moore, The MathWorks

This paper discusses the various stages of architectural design for embedded systems, from the design of complex algorithmic components through to the deployment of these components on a component-based platform such as AUTOSAR.

Alan Moore is an Architecture Modeling Specialist at The MathWorks. Alan has 22 years experience in the development of real-time and object-oriented methodologies and their application in a variety of problem domains. He is an active member of the Object Management Group (OMG) and was a long time cochair of OMG’s Real-time Analysis and Design Working Group. Most recently he acted as the Specification Architect for the SysML specification and has worked closely with the Modeling and Analysis of Real-Time and Embedded Systems (MARTE) specification team.

AUTOSAR-Compliant Functional Modeling with MATLAB, Simulink, Stateflow, and Real-Time Workshop Embedded Coder of a Serial Comfort Body Controller

Andreas Köhler, Volkswagen

Volkswagen AG and further partners are developing together a fully functional Body/Comfort ECU for a Volkswagen series-production vehicle that will be furnished with Automotive Open System Architecture (AUTOSAR) compatible software. The aim of the project is to check, over a period of 12 months on a day-to-day basis, how the demands of the automotive industry with regards to real time functionality can be met.

Volkswagen will also evaluate software integration, migration opportunities, use of resources, software runtimes, quality of specifications, and cost. One goal of this project is to integrate functions already developed in MATLAB and Simulink into this Body/Comfort ECU. In cooperation with The MathWorks, the tool chain has been prototypically extended so that the functions could be generated by means of the code generator interface of the AUTOSAR platform. Then the AUTOSAR Control Unit can be mounted in a Passat sedan, thus creating an authentic test environment.

In this paper we point out how Volkswagen will use the model-based function development in the AUTOSAR context.

The Role of Real-Time Workshop Embedded Coder in Supporting the Vision of Cummins for Model Based Developement

Mark W. Pyclik, Cummins Inc.

Cummins designs and develops software using the Product Line Architecture (PLA) approach. This approach has served the company well in meeting the many customer needs that require changes in the software. Although PLA gives Cummins advantages in delivering high-quality software, it also creates challenges in realizing Model Based Development (MBD) as the vision for developing controller software, due to the higher complexity and open interface designs leveraged within PLA. For the last decade, Cummins has been vigorously pursuing MBD, with the major challenge of finding an automatic code generation tool that could generate embedded code within the constraints of PLA. Cummins already uses automatic code generation for its current embedded controller development. Last year, after an extensive two year pilot program with The MathWorks, Cummins selected Real-Time Workshop Embedded Coder for its future controller development efforts.

This paper describes how Real-Time Workshop Embedded Coder is fulfilling the future needs of Cummins as its next-generation automatic code development platform.

Production Code Generation with Model-Based Design

Tom Erkkinen, Embedded Application Manager, The MathWorks

This talk describes Model-Based Design with automatic code generation for production ECUs. It demonstrates new Simulink and Real-Time Workshop Embedded Coder technologies involving code optimization, software integration, and processor-in-loop verification. The presentation first introduces Simulink code generation to novices, and then highlights important new capabilities for code generation experts. Industry examples are also discussed.

Tom Erkkinen leads initiatives to foster industry adoption of production code generation for embedded applications at The MathWorks. Before joining The MathWorks, Tom worked at Lockheed developing real-time software for missile systems and embedded software for the Space Shuttle Robotic Manipulator System. Tom also helped develop a variety of HIL test labs at NASA Johnson Space Center. He has worked at commercial companies developing and implementing safety-critical code generation and automatic unit test tools in the aerospace and automotive sectors. Tom holds a B.S. degree in Aerospace Engineering from Boston University, and an M.S. degree in Mechanical Engineering from Santa Clara University.

Development of a Production Adaptive Cruise Controller for Heavy Trucks Using Model-Based Design and Production Code Generation

Magnus Eriksson, Kristian Lindqvist, Håkan Andersson
Scania CV AB, Dept: Systems & SW

The Scania ACC system was developed using Model-Based Design and code generation. One hundred percent of the production application code for the longitudinal controller was automatically generated from Simulink and Stateflow models.

The MATLAB tool chain was used for modelling of the control algorithm and code generation. The tools also supported the test process with model coverage measurements. Simulink, Stateflow, and Dymola were used in cosimulation for plant modelling. These models were used to develop the control algorithms and evaluate the overall fuel efficiency of the system.

The generated code was examined automatically to check MISRA compliance and manually to ensure that it met the requirements of Scania. A tool that generated test vectors with high MC/DC coverage was developed, and the test vectors were used to check that the model output was the same as that of the generated code running on the target hardware. It was proven that the same system behaviour could be expected from the simulation environment and from the automatically generated code running on the actual target hardware.

These development methods and tools enabled the production of higher quality, more efficient functionality than with previously employed methods within Scania.
This paper gives some examples of experiences gained by using Model-Based Design and production code generation for a heavy truck ACC application.

AVS: A Test Suite for Automatically Generated Code

Ekkehard Pofahl, Ford Motor Company

As automatically generated code is planned for use in highly safety-relevant automotive electronics (such as for chassis, brakes, and vehicle stability control), Ford of Europe and Continental Automotive Systems saw a necessity to provide measures that would ensure at least the same safety integrity for automatically generated code as is required for hand-coded electronic control units (ECUs), mostly using C as the programming language.

In order to validate the quality and robustness of the code generator as well as the downstream tool chain (for example, the compiler and linker), they developed the Automotive Code Validation Suite, or AVS, for automatically generated code. This suite is employed to check transformation from the graphical design down to the final executable on different target platforms. Its processor-in-the-loop (PIL) and target-in-the-loop (TIL) testing capabilities permit the actual target hardware (ECU) or a target-like hardware (evaluation board) to be incorporated into the validation. The versatile architecture of the AVS is ideally suited for optimization-level validation (OLV) and regression testing. AVS allows easy adaptation to various tool chains and hardware environments (for example, the code generator, compiler, linker, and microcontroller).

New Techniques for Model Verification and Validation

Brett Murphy, The MathWorks

In production programs, design errors are very costly, and in safety-critical systems there is zero tolerance for design errors. In these programs verification, validation, and test (VV&T) are critical.  The demands of these programs drive the need for tools and process to enable rigor and automation in VV&T for Model-Based Design.  The MathWorks has developed a set of Best Practices that, when adopted, can help find errors early before they become costly problems.  This talk will present these, as well as new tools The MathWorks has released to make these Best Practices easier to implement.

Brett Murphy is responsible for the technical marketing of verification, validation, and test products at The MathWorks. Brett has extensive experience in controls analysis, real-time software development, and systems engineering in the aerospace and embedded systems industries. Brett worked for two years at Fujitsu in Tokyo on advanced robotics. He then worked for Space Systems/Loral (SS/L) in Palo Alto, CA. At SS/L Brett was a lead engineer, developing hardware-in-the-loop satellite simulators, complex real-time systems that simulate the complete operation, communication, and environment of a satellite in orbit. In 1995, Brett joined Real-Time Innovations, Inc. (RTI) and guided the marketing of RTI’s embedded software test tools and network middleware for 10 years. Brett holds a B.S. and M.S. in Aerospace Engineering from Stanford University.

Spreadsheet-Based Model Test

Weiqian Sun, Ford Motor Company

The delivery of high quality embedded software requires consistent and robust testing throughout the software development lifecycle. The development and reuse of tests across virtual rapid prototyping and hardware test environments is a fundamental capability that has typically been managed with multiple manual translation steps. This paper describes a pragmatic approach toward test case generation and reuse that uses a combination of Simulink, Stateflow, and Excel spreadsheet extensions.

The paper describes and presents key production software testing results generated using the integrated Simulink, Stateflow, and Excel test design environment. It illustrates how our engineers were able to efficiently construct tests and verify their feature designs using intuitive test-build and visualization environments. It also provides a description of how the Model Coverage Tool in Simulink Verification and Validation from The MathWorks was used to speed the test development process and achieve high quality designs.

Cold Engine Emissions Optimization Using Model-Based Calibration

Clive Tindle, General Motors - Holden Ltd

Emissions calibration development of gasoline engines is becoming increasingly demanding for the calibration engineer, due to a number of factors: time for development programs, lower exhaust tailpipe emissions, and reduced engine-out emissions that take full advantage of precious metal cost reductions in the catalyst. To meet this demand, the technological complexity of the gasoline engine has increased with the introduction of continuously variable intake and exhaust camshafts and, more recently, the introduction of fuel injection strategies on direct injection engines, such as double injection. With this technology, engine data generation and calibration optimization must be handled using a model-based approach and design of experiments (DOE).

This paper focuses on the application of design of experiments methods and optimization of two-stage statistical response models to develop an engine calibration to current worldwide standards, using MATLAB and Model-Based Calibration Toolbox. The model-based methods used have been shown to be capable of producing calibration values for the main actuators. Examples are presented which relate to recent applications during vehicle development. These show that the use of model-based methods is no longer a luxury, but a necessity in engine calibration.

Steady-State Engine Modeling for Calibration: A Productivity and Quality Study

Ken Butts, Toyota Engineering and Manufacturing, NA

Product development at Toyota Motor Corporation is constrained by the engineering time and by the resources required for engine development. To mitigate this constraint, Toyota is seeking ways to improve engine development productivity. In this paper we present one such effort to improve engine calibration productivity. The paper describes a steady-state engine modeling process that leverages Design-of-Experiment-based (DoE) statistical modeling and engine test-bench automation. We also provide quantitative assessments of DoE-based test measurement time, command-line driven statistical modeling, and engine model quality.

The MATE Approach: Enhanced Simulink and Stateflow Model Transformation

Ingo Stürmer, Model Engineering Solutions

In this paper we present the Model Advisor Transformation Extension (MATE). The purpose of MATE is to complement the functionality of the MathWorks MATLAB, Simulink, and Stateflow, Model Advisor, and to extend the tool’s capabilities with regard to model transformation and improvement functions. Examples of MATE features are: automatic or interactive model analysis and repair functions; design pattern instantiation; beautifier operations. We present typical use cases for MATE and discuss the relevance of the MATE approach compared with other available tools and approaches.

Methods for Verification and Validation

Goran Begic, The MathWorks

Verification and validation is critical for the implementation of Model-Based Design in production programs. This master class introduces new concepts and tools for effective verification and validation based on model-analysis techniques. Participants learn how to build component test environments for model and code, how to ensure completeness of tests, and how to automate test execution and reporting through series of examples based on best practices in model verification and validation.

Goran Begic is a product marketing manager at The MathWorks, working on tools for verification, validation, and testing of embedded systems. He has over seven years of experience in technical marketing, sales, and support of software development and testing tools. In 1996, he earned a B.S. in Electrical Engineering from the University of Zagreb.

Production Code Generation

Steve Toeppe, The MathWorks

Steve Toeppe is a development manager on the Real-Time Workshop Embedded Coder team, which develops features for Real-Time Workshop Embedded Coder and Model Advisor, and provides support to customers with production deployments of these tools. Steve joined The MathWorks in 2000. He has 27 years of industry experience developing embedded systems for powertrain control systems, combat control systems, robotics, computer vision, and industrial automation. Steve worked at Ford Motor Company for seven years as supervisor in powertrain controls systems and as team leader for software technologies at Ford Research Laboratories. While at Ford Research, he was actively involved in defining specifications for code generation technology, preparing the MAAB Style Guide, and authoring numerous Model-Based Design technical papers. Previously, Steve worked at General Dynamics Land Systems and several robotics and computer vision companies. He received an M.S.E.E. from Wayne State University (1988) and a B.S.E.E. from the University of Michigan (1980).

Advanced Programming Techniques in MATLAB

Loren Shure, The MathWorks

This master class covers two important MATLAB topics: handling memory efficiently and choosing among the rich set of function types. Participants gain an understanding of how different MATLAB data types are stored in memory and how to program in MATLAB to use memory efficiently. Recent versions of MATLAB have introduced several new programming concepts, including new function types; the class illustrates and explores the usage and benefits of the various function types under different conditions. Learn how using the right function type can lead to more robust and maintainable code. Demonstrations will show how to apply these techniques to problems that arise in typical applications.

Loren Shure has worked at The MathWorks for more than 20 years. She has co-authored several MathWorks products in addition to adding core functionality to MATLAB. Loren currently works on the design of the MATLAB language. She graduated from MIT with a B.S. in Physics, and from the University of California, San Diego, Scripps Institution of Oceanography with a Ph.D. in Marine Geophysics.

Modeling Multidomain Physical Systems in Simulink

Steve Miller, The MathWorks

Steve Miller joined the Design Automation Marketing group at The MathWorks in 2005. Prior to that, Steve worked at Delphi Automotive in Braking Control Systems and at MSC.Software Adams consulting in various capacities at Ford, GM, Hyundai, BMW, and Audi. Steve has a B.S. in Mechanical Engineering from Cornell University and an M.S. in Mechanical Engineering from Stanford University.