Inspiring Automobiles: Electronic Systems as Significant Contributors and the Challenge of Dealing with Complexity

Armin Müller, Director Innovations and Concepts at Dr. Ing. h.c. Porsche AG

In his keynote address, Armin Müller will give an inside view into the development of inspiring automobiles by talking about perceptible applications from a sports car manufacturer’s point of view. He will discuss approaches how Porsche will deal with the complexity of the digital development. Armin Müller will talk about the challenges related to the development process of electronic systems and how this process fit into the complete vehicles development process.

Armin Müller is Director, Innovations and Concepts, at Dr. Ing. h.c. F. Porsche AG, a position he has held since 2006. He held several positions at Daimler Benz where he started 1986 in the development of brake and traction control systems and in 1990 he became the team lead for traction control development. In 1992 he assumed the project management position for ESP. The development of the first ESP system was successfully finished in 1995. Between 1995 and 2000 he was head of several departments: Advanced Development Vehicle and Systems, Advanced Development Driving Systems, and Series Development Vehicle Dynamics. In 2000 Mr. Müller became Director of Development at ZF Lemförder, where he also was member of the division management board and responsible for the worldwide development. 2003 he organized the establishment of ZF Engineering. Armin Müller received his diploma degree in mechanical engineering 1989 from the University of Stuttgart.

ASAM-MBFS: A Standardized Block Library as Enabler for Efficient Model-Based Collaboration

Dr. Thomas Burger, Continenta and Johann Gabler, Audi AG

Within the automotive industry, Model-Based Development based on MATLAB®, Simulink®, and Stateflow® has established itself as a quasi-standard. One important aspect of this method is the possibility of close cooperation between OEM and supplier throughout the complete development cycle: Starting in early requirement phase via requirement specification given by an executable model up to early validation in the vehicle via rapid prototyping (RPT).

The backbone of this collaboration is the common understanding of the models and the ease of their exchange. Within the ASAM-MBFS (Model-Based Function Specification) working group, a standard for an automotive block library has been specified. In this context, Continental and Audi AG together drove the specification of an implementation based on MATLAB and Simulink. A corresponding reference block library has been worked out and is provided via ASAM. This talk gives an overview, from the motivation to the implemented block library.

Dr. Thomas Burger (Continental) studied physics at the University of Regensburg. There he made his doctoral thesis in biophysics at the working group Computational Intelligence and Machine Learning of Prof. Dr. Lang from 1996 to 1999. Thereafter, he joined the automotive business unit of Siemens AG and worked on the software development of engine control units. Currently, he is project manager within the Powertrain Engineering Groups of the Continental division Powertrain. He is responsible for the topic Rapid Prototyping of embedded in System Design Automation (SDA), which is representing the generic environment at the division Powertrain for the model-based function development based on MATLAB and Simulink.

Johann Gabler (Audi AG) studied electrical and information engineering at the technical college at Munich. From 1994 to 1998 he worked for DaimlerChrysler in functional and software development for chassis control systems. Thereafter he was participated significantly in the advanced product development for steer- and brake-by-wire systems at TEMIC Automotive in Nuremberg. Since 2001 he has worked at AUDI AG in Ingolstadt in the department of electronic control units for powertrain systems and takes the responsibility for model-based function development and rapid prototyping systems based on MATLAB, Simulink, and Stateflow.

Developing AUTOSAR Software Components with Model-Based Design

Dr. Joachim Schlosser, The MathWorks

The development of AUTOSAR Software Components requires interaction with Authoring tools used to develop a vehicle's architecture and ECU topology as well as RTE generation environments.  This master class demonstrates how the development of AUTOSAR Software Components based upon Simulink and Real-Time Workshop® Embedded Coder™ is incorporated into a workflow using these tool environments.

Functional Variants Handling in Simulink Models

Wojciech Przystas, Daimler AG R&D

Today’s cars are characterized by a multitude of functional variants. There are numerous sources of variability – starting with the adaptation to diverse markets, ending with offering different technical features, and thus, different ECUs (Electronic Control Units). Inevitably, this is reflected in model based function development using Simulink.

Simulink models describe, in particular, the functional algorithm of real time systems. In practice, the aspects of modeling and configuration of functional variants overlay the “basic” algorithm. This happens in context of creation of reusable and configurable functional modules. These two aspects, the modeling and configuration, lack the necessary means of variability description, and thus cannot be unambiguously described through a reliable process. As a consequence, a reliable classification by the modeler, or the code generator, is not always possible.

There is a demand for concepts, which systematically describe variability handling in Simulink, for example.:

• Which information concerning basic blocks must be available to describe functional variants?
• Where and how can this information be stored?

This article describes an approach for handling functional variants in Simulink models. It is based on the separation of general and block-specific variability information, thus allowing:

• A uniform description (and therefore a uniform configuration) of functional variants
• Variant-specific presentation/visualization of these variants in Simulink models (showing the difference between “regular” and “variant specific” blocks), as well as the recognition of dependencies between the variants

This approach was developed within a research project and is currently being applied in a series project in the development of Mercedes-Benz® cars. Its scope comprises the implementation of a variant block set, a variant specific API (Application Programming Interface) based on MATLAB functions, as well as a configuration tool.

Wojciech Przystas studied computing science and computational linguistics at the University of Stuttgart. Since 2007 he has been with Daimler AG R&D, working as a Ph.D. student in the area of variant configuration of model based embedded software.

Introduction to Practical Uses of the Plant Model in Model Based Development

Takayuki Kubo, Aisin AW Co., Ltd.

Our applications of Model Based Development are introduced. The first section will cover activities for HILS development and Controller design. The second section explains the practical development of the plant model – an automatic transmission model – with SimDriveline™. The same example will be used for applying Simulink® Parameter Estimation™ to analyze this model.

Takayuki Kubo is a senior specialist in the Advanced Engineering group of Aisin AW Co., Ltd.

System-Level Design of Electrohydraulic and Mechatronic Systems

Steve Miller, The MathWorks

Responding to climbing fuel prices and environmental concerns, automakers and suppliers are scouring vehicles looking for ways to improve efficiency without sacrificing performance. Advances in mechatronic systems have created an efficient alternative to electrohydraulic systems, and companies are investigating those alternatives and replacing hydraulic pumps and cylinders with power electronics and electric motors. Due to the complex nature of these systems, it is critical to analyze them at the system level to understand the control and safety critical requirements of these systems. This talk focuses on how simulation can help in the analysis and design of electrohydraulic and mechatronic systems using the specific example of power steering systems.

Steve Miller is responsible for the technical marketing of products for modeling physical systems at The MathWorks. Steve has extensive experience in controls analysis, having developed algorithms for braking control systems at Delphi Automotive. After doing algorithm development for 2.5 years using MathWorks™ products, Steve then worked as a consultant for MSC.Software doing multi-body simulation at a number of automotive companies in the United States (Ford, GM, Bosch) and in Germany (BMW, Audi) for five years before joining The MathWorks in 2005. Steve holds a B.S. in Mechanical Engineering from Cornell University and M.S. in Mechanical Engineering from Stanford University.

Practical Application of Model-Based Control for ACC

Ian Blake, Jaguar

Adaptive Cruise Control (ACC) offers a considerable non linear control challenge, as the driver expects different actions dependent on the driving context. The control algorithms can be realized in various ways; maps are often favored for their flexibility and easy processing. However, this flexibility can make definition and editing difficult.

This paper shows the use of fuzzy logic techniques in MATLAB to generate and edit the map surfaces in a consistent and smooth manner; the use of Model-Based Design for control autocoding, simulation, and test is also discussed, and possibilities for future optimization work using toolboxes is also considered.

Ian Blake has been working on Adaptive Cruise at Jaguar Cars for 10 years, including the production release of ACC into the market. In that time he has been involved in design, development and testing of the system, and is currently pursuing further refinement and wider fitment of the system. In 2003 he completed an M.Sc. at Coventry University relating to ACC control using fuzzy logic.

Prototyping and Deployment of Real-Time Signal Processing Algorithms for Engine Control and Diagnosis

Fabrice Guillemin, IFP - Institut Français du Pétrole

Future internal combustion engine technologies require an accurate combustion monitoring and control. This can be performed through high frequency recordings and processing of combustion related variables such as cylinder pressure or knock signals. In the context of innovative combustion management studies, IFP develops flexible platforms in order to prototype signal processing algorithms on test bench or vehicle, and then to deploy them on DSP based embedded targets.

Description of the IFP platforms:

• Prototyping platform is an xPC Target™ PC with specific engine synchronous timer board and acquisition hardware. This board and corresponding drivers for Real-Time Workshop® have been developed at IFP. Combustion Analysis and knock signal processing algorithms are developed in Simulink with Signal Processing Blockset™.
• Deployment platform is a high performance integrated system based on the TI-C6727 DSP. This "On Target" facility is a valuable asset to bring validated algorithms into an embedded industrial device and increase processing capability at a low energy consumption.

The code generation and build process uses:

• Simulink
• Signal Processing Blockset
• Real-Time Workshop Embedded coder
• Target Support Package™ TC6
• Embedded IDE Link™ CC

Thanks to this tool chain and TI C67x Target Function Library optimized code, the deployment platform updates online algorithms computations, engine synchronously, at high resolution (0,1 crank angle degree), for multiple channels (up to 8).

• Research engineer in Engine Control and Signal Processing department at IFP since 2005
• Consultant engineer at ATERSIM, France in 2005
• Consultant and training engineer at The MathWorks, France, from 2000 to 2004

Production Code Generation Time Machine

Tom Erkkinen, The MathWorks

This talk describes Model-Based Design with automatic code generation for production ECUs. It demonstrates how Simulink and
Real-Time Workshop Embedded Coder have continuously improved throughout the years in key technology areas such as code optimization, fixed point design, software integration, and code verification. The presentation will interest code generation novices as well as experts interested in latest features and capabilities. 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.

PHEV Control Strategy Optimization using MathWorks Distributed Computing Tools: From Pattern to Tuning

Sylvain Pagerit, Argonne National Laboratory

Plug-in Hybrid Electric Vehicles (PHEVs) offer a great opportunity to significantly reduce petroleum consumption.  However, their fuel economy varies highly pending on their All-Electric-Range (AER) and the distance traveled which makes an optimized control strategy difficult to implement. Using the Powertrain System Analysis Toolkit (PSAT) developed in MATLAB, Simulink, and Stateflow, we designed a global optimization algorithm to extract control patterns and implement them in PSAT. We then used a Dividing RECTangle (DIRECT) algorithm adapted to PSAT to tune the parameters of the new control. The amount of computation needed to run these algorithms lead us to use distributed computing for both of them. The MathWorks distributed computing tools allowed us to distribute these algorithms very easily with very minimal coding. These controls are now part of the new release of PSAT.

Model-Based Development: Realizing Fully Integrated Algorithm and Software Development for Production Automotive Electronic Control Units

Peyman Moradshahi, American Axle & Manufacturing

American Axle & Manufacturing (AAM) is utilizing an advanced Model-Based Development (MBD) process for several of its production programs.  These programs are using S12 and S12X fixed-point microcontrollers from Freescale.

It has generally been established that Model-Based Development can provide dramatic improvements in productivity over traditional hand-coding.  Productivity improvements have resulted from the powerful combination of modeling and simulation with Simulink and the generation of production code with Real-Time Workshop Embedded-Coder from Simulink models.  However, a bottleneck remains that hampers the full productivity improvement potential.  The code generated from modeled subsystems still has to be integrated and tested by hand with the BIOS code which includes scheduling, data conversion and device driver code. 

This paper describes the Model-Based Development process used within AAM to solve this bottleneck, resulting in significant further improvements in productivity and product quality.  In particular, AAM performs full integration and test of the control system and BIOS within Simulink before any code is generated. The AAM solution is enabled by a combination of the MathWorks tools together with third party tools for Simulink from SimuQuest. Ultimately production code for the entire software application is automatically generated with Real-Time Workshop Embedded-Coder.

Peyman Moradshahi is engineering manager at AAM and oversees all electronically integrated production programs. He has developed embedded Windows® and Unix® based software in various programming languages in both automotive and medical industries. He has utilized MATLAB and Simulink for control algorithm development throughout his career. He holds B.S. and M.S. degrees in Electrical Engineering.

Enriching the AUTOSAR Component Model

Prof. Werner Damm, Chair for Safety Critical Embedded Systems at the Carl von Ossietzky University of Oldenburg: .

Starting from the key benefits and major challenges the AUTOSAR Design Methodology offers Prof. Damm is going to discuss building blocks of a design methodology for distributed real-time automotive applications striving to reconcile the advantage of early system-level analysis with the overall AUTOSAR objective of decoupling function design from its implementation.

He will present an approach to conservatively extend the AUTOSAR component model towards what he calls rich component interface specification, where “richness” refers to three dimensions namely the capability to express the multitude of non functional constraints, sufficient expressive of interface specification language, and contract-based interface specifications, allowing in particular using so-called vertical assumptions for capturing resource requirements at system-level.

Prof. Werner Damm holds the Chair for Safety Critical Embedded Systems at the Carl von Ossietzky University of Oldenburg. He is member of the Board of Directors of OFFIS, the Chairman of the SafeTRANS competence cluster, integrating leading companies and research institutes in the transportation domain, Chairman of the Steering Boards of EICOSE (European Institute for Complex Safety Critical Systems Engineering), the Artemis Innovation Cluster on Transportation, and member of the ITEA2 Roadmap3 Steering Board.

His recent research covers foundational research on mathematical models of embedded systems, specification languages, hybrid systems, formal verification methods, and real-time and safety analysis. This is complemented by applied research with industrial partners in avionics, automotive, and train system application. The focus of this research is on enhancing model-based development processes with formal method-based approaches to verification, testing, and safety and real-time analysis, as well as on enabling component-based design for embedded systems.

Model-Based Design for Safety Critical Applications

Mirko Conrad, The MathWorks

This talk discusses how Simulink and Real-Time Workshop Embedded Coder can be leveraged to develop and certify safety critical automotive applications. It covers the mapping of IEC 61508 objectives onto Model-Based Design plus demonstrates recent Simulink® Verification and Validation™ product support for automotive safety standards.

Real-Time Control and Integration of a High-Fidelity Driver-in-the-Loop Motion Platform

Julian Hodgson, Red Bull Technology

By any measure Formula One racing is largely concerned with the issue of driving a vehicle at, or near, its performance limit. This raises important questions about extending the performance limit while being mindful of the fact that these improvements must be achievable by a driver who closes the feedback loop. The cost of track testing, regulatory limitations on the number of testing days, and the need for rapid turnaround means that virtual prototyping is becoming
increasingly important. In order to extract the maximum benefit from this it is necessary to keep the pilot in the loop since the driver forms a critical component in the overall feedback system.

To be useful this requires a very high performance motion system, as well as graphics and sound generation systems to cue the driver. The high accelerations seen in Formula One, and the fact that a professional driver operates the vehicle at the performance limit, makes the task of appropriate cueing particularly challenging.

A key part of the project is a distributed real-time simulation and control environment. This must provide simulation of the vehicle dynamics, control of the motion platform, and integrate with the visual system and existing off car systems.

Because of the requirement for a distributed solution that can integrate with existing off car systems Red Bull Technology are working with McLaren Electronic Systems Limited and Speedgoat GmbH in developing a new Rapid Prototyping environment, all built on simulation and code generation tools from The MathWorks.

Following a brief overview we will describe the distributed hardware/software architecture and the technical challenges involved in delivering this critical part of a simulator.

Panel Discussion: Best Practices for Establishing a Culture of Model-Based Design Within Automotive Organizations

The transition to Model-Based Design requires careful management to enable the full realization of its benefits. Establishing a culture that embraces the process is key to achieving the associated time and cost savings. In this panel discussion, representatives from leading automotive companies will discuss their experiences in establishing Model-Based Design within their organizations. Panelists will discuss:

• Their organization’s progress in adopting a culture of Model-Based Design
• How they measure the impact of Model-Based Design on the organization
• The lessons they have learned – both positive and negative – in transitioning to Model-Based Design

Quality Improvement by Continuous Usage of PolySpace™ Products with Model-Based Design in the Area of Safety Systems

Dr. Almut Hochstädter, Conti Temic Microelectronic GmbH and Mr. Jochen Retter, Conti PolySpace paper

Continental Passive Safety now uses a fast and secure front-loaded development process for safety critical software. A main key inside this development method is the systematic usage of PolySpace products for all existing and upcoming projects. PolySpace is also the key tool for the validation of 3rd-party Software being integrated into an airbag ECU, regarding not only AUTOSAR-compliant components. For upcoming projects, a part of Airbag Software will be modeled in Simulink and Stateflow. PolySpace™ Model Link™ SL will be the key tool for validation of these models and the resulting generated code. Continental reveals in this talk a seamless approach of a front-loaded development using MATLAB, Simulink, Stateflow, and PolySpace products for production intend airbag ECU software.

Generating Hardware Descriptions from Automotive Function Models for an FPGA-Based Body Controller: A Case Study

Matthias Traub, Daimler AG

Today's innovations in the automotive sector are, to a great extent, based on electronics. New features are often based solely on embedded electronic devices. Therefore, developers have to face higher demands in the field of electric/electronic architectures. Regarding aforementioned functions a lot of more computational power has to be integrated into the car, while automotive constraints like reliability, minimum costs and real time capability still have to be fulfilled.

Assistant systems, engine management, and telematic-applications are one of the main drivers for this trend. To fulfil the requirements new chip-architectures including multicore-systems, Field Programmable Gate Arrays, and Structured ASICs, give an interesting alternative. The latter offer more freedom regarding a customized hardware-software partitioning that better fits the requirements of performance and flexibility. To handle the partitioning task, novel methodologies as well as appropriate and transferable description languages are necessary. Estimations on system performance, reliability and cost must be available at very early stages in order to achieve efficiency. Automatic transformation of existing C or C++ based solutions into hardware descriptions is feasible. However, the performance benefit is insignificant compared to the resources spent. The main reason for this is the different nature of programming languages being sequential and concurrent hardware statements that allow parallelization. Model-Based Design introduces a higher level of abstraction, which typically is independent from any hardware or software decision. Therefore, the same model can be used for hardware or software implementation. This allows a better design-space exploration regarding the partitioning and may lead to better solutions.

Today more and more applications are developed using Model-Based Design in the automotive domain.  While C code generation is already used in series development, no satisfying solution for VHDL® or Verilog® generation was available on the market for Simulink and Stateflow models. Simulink® HDL Coder™ from The Mathworks introduces the ability to divide model-based functions in hardware and software modules for acceleration of time critical applications.

A case study for a FPGA-based body controller including automatically generated HDL code will be presented in this paper. Furthermore, the advantages and open issues will be discussed. The conclusion gives an outline to an application scenario of this technology for AUTOSAR.

Early Verification and Validation in Model-Based Design

Amory Wakefield, The MathWorks

This presentation will introduce some best practices for repeatable and exhaustive verification in the Simulink environment. It will discuss how early verification and validation in Model-Based Design can improve the overall quality of products. Topics will include modeling standards, model testing, and model coverage.  It will also show how to prove the correctness of generated code using PolySpace code verifiers.

Using Model-Based Calibration Toolbox™ Multimodels for Cycle-Optimized Diesel Calibration

Joshua Styron, Ford Motor Company

Modern diesel engines have many degrees of freedom that must be simultaneously adjusted to optimize efficiency, emissions, and performance. Due to the complexity of the interactions amongst the input parameters, it is not possible to decompose the calibration process into a set of smaller, simpler models. Further complicating the calibration process, there is a strong desire to make tradeoffs amongst different speeds and loads to meet various cycle-based emissions and fuel economy targets. Simply building models with two more inputs to represent the entire engine map may result is a loss of fidelity as input parameters may interact differently across the engine operating range. Model-Based Calibration Toolbox can now generate multimodels which essentially selects the best local model at each speed and load operating point. These global models retain all of the fidelity of the local models, while allowing real-time tradeoffs at discrete speed and load points in cycle optimizations.

In CAGE, toolbox multimodels were used to generate steady-state calibration set points for an entire engine map that minimize fuel consumption while simultaneously meeting cycle emissions and global NVH targets.  In addition, many constraints were utilized to assure the calibration was realistic.  Map gradient constraints helped to assure the resulting optimized maps would be smooth enough.  Boundary constraints were used to limit the optimization to attainable points, while models of critical engine limits like peak cylinder pressure and turbine inlet temperature assure the calibration remains within the critical engine limits.

Joshua Styron received his Ph.D. from the University of Illinois at Urbana-Champaign where he developed a novel laser-diagnostic technique to quantify vapor fuel distributions of multicomponent fuels inside an optically accessible, spark ignition engine. This technique was used to minimize hydrocarbon emissions during cold-start conditions due to poor mixture preparation. Upon graduation, Joshua was hired by Ford Motor Company for a brief rotation through Powertrain Systems Engineering.  He then moved on to Ford Research, where he invented and developed variable compression ratio engine mechanisms as well as sensing concepts, control strategies and calibrations. Joshua then moved on to Diesel engine development where he worked to develop a new combustion system starting from CFD and proceeding through single-cylinder engine testing and multicylinder engine validation.  He is currently leading mapping efforts for Ford's North American Diesel engine programs to optimize steady-state calibrations.

Early Verification and Validation Within Model-Based Design

Dr. Marc Segelken, The MathWorks

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

Developing AUTOSAR Software Components with Model-Based Design

The development of AUTOSAR Software Components requires interaction with Authoring tools used to develop a vehicle's architecture and ECU topology as well as RTE generation environments.  This master class demonstrates how the development of AUTOSAR Software Components based upon Simulink and Real-Time Workshop Embedded Coder is incorporated into a workflow using these tool environments.

Modeling Multidomain Physical Systems in Simulink

Steve Miller, The MathWorks

This master class focuses on the use of MathWorks physical modeling products in doing system-level design of electrohydraulic and mechatronic systems. During this session, the example shown in the main conference track (power steering system) will be used to illustrate how to build models for use in doing tradeoff studies and control development.

Steve Miller is responsible for the technical marketing of products for modeling physical systems at The MathWorks. Steve has extensive experience in controls analysis, having developed algorithms for braking control systems at Delphi Automotive. After doing algorithm development for 2.5 years using MathWorks products, Steve then worked as a consultant for MSC.Software doing multibody simulation at a number of automotive companies in the United States (Ford, GM, Bosch) and in Germany (BMW, Audi) for five years before joining The MathWorks in 2005. Steve holds a B.S. in Mechanical Engineering from Cornell University and M.S. in Mechanical Engineering from Stanford University.

Code-Based Verification using PolySpace™ Products

Marc Lalo, The MathWorks

Minimizing the risk that errors make it into released software applications is a key objective for the development of embedded software.  This master class will show you how code-based verification with PolySpace products addresses this goal by checking coding standards and proving the absence of errors for all instructions in the source code. We’ll use various examples – source code and models – to illustrate how code-based verification fits within hand-written development processes as well as Model-Based Design with automatic code generation.

Power tips for Using Simulink on Large Projects

The MathWorks

In this class, we will examine industry best practices for major capabilities introduced in recent years that support using Simulink for large-scale applications. Practical examples and industry workflows will be emphasized. Key topics include:

• Using model referencing and busses with component-based modeling techniques
• Performing version control and configuration management
• Applying model patterns that yield clear designs and generate efficient code
• Managing large data sets associated with your model and code
• Running simulations as efficiently as possible
• Adapting Simulink to your environment, including customizing user interfaces and automating model guideline checks

Design to Hardware: A Case Study on Implementing a Knock Detection Algorithm

The MathWorks

This session will demonstrate how MathWorks tools can be used to design, implement, and test an algorithm in the context of an engine knock detection and correction application. The demo will provide an overview of the design process of the fixed-point algorithm (including filters and control logic), as well as specify test cases to verify correct behavior. The algorithm will be implemented on a DSP using C code generation as and on an FPGA using HDL code generation. This workflow can be extended to other control or signal processing applications.