Tuesday, 17 April, Keynote Presentation
Dr. Andreas Titze, Volkswagen
Model-Based Design and AUTOSAR are established tools in automotive software development. However, critics continue to blame both paradigms for the increasing demand in resources. This keynote presentation gives some answers to the question of managing complexity from a Volkswagen point of view.
Tuesday, 17 April, Customer Sessions
Hessel van Dijk and Dr. Michel Paas, Ford Motor Company
Nadja Wirth, TECOSIM GmbH
In the vehicle development process, tradeoffs between weight and cost, as well as performance requirements involving crash safety, NVH, vehicle dynamics, durability, and other attributes, need to be established. Multidisciplinary design optimization (MDO) is a key enabler to solve these complex tradeoff studies. At Ford Motor Company, MDO has been integrated into the global product development process. To comply with all performance, weight, and cost targets and to deliver reliable and robust design proposals in time, a toolset for automated cross-attribute optimization based on MATLAB has been developed. The toolset has been deployed successfully throughout various vehicle programs.
This presentation elaborates on the automated data generation functionality, unique data-mining and meta-modeling features, multiobjective optimization algorithms, and data visualization techniques.
In addition, a MATLAB based graphical user interface is discussed.
Hasan Uzun, Nataša Kieft, Christian Manz, and Steffen Waldmann, Mercedes-AMG GmbH
Modern engine calibration engineers face the challenge of meeting both fast-changing regulations on emission restrictions and increasing requirements on performance. Furthermore, while development and test times need to continually be reduced, the number of parameters to be optimized is increasing drastically as new advanced technologies become available.
To manage the number of calibration functions and parameters in a structured way, Mercedes-AMG developed the in-house AMG Automation Tool to meet the engine calibration challenges. The tool makes use of MATLAB, Model-Based Calibration Toolbox™, and Parallel Computing Toolbox™ to enable AMG calibration engineers at all levels of expertise to extract the highest possible engine performance while minimizing exhaust emissions.
Dipl.-Ing. Clément Val, CEESAR
Dipl.-Ing. Jörg Küfen, RWTH Aachen, Institut für Kraftfahrzeuge
The increasing number and complexity of advanced driving assistance systems require an ever-increasing amount of field data. Data corresponding to the actual use of vehicles by ordinary drivers in real driving conditions is needed to calibrate new systems, identify their shortcomings and evaluate their impact. Although necessary, experiments in controlled conditions, such as test-bench, test-track, or driving simulator, only allow observations in a finite number of predefined and generally simplified scenarios.
In this session, we show how we have implemented, using MATLAB, a very powerful framework that automates data management and processing and completely separates technical tasks from more scientific ones. Using this software, analysts have all the necessary tools to browse and search for specific things in data, transform the logged records into usable results with their own algorithms, and visualize and annotate everything (original signals, derived data, geographical position, video, etc.) in a single interface.
Axel Grobe and Abdulvahap User, IAV GmbH
Matthias Faustmann, Volkswagen AG
The use of Model-Based Design for developing AUTOSAR-compliant electronic control units (ECUs) at Volkswagen began more than five years ago. At the MathWorks Automotive Conference 2007, Volkswagen presented a proof-of-concept project “KSG on AUTOSAR.” That presentation illustrated the integration of AUTOSAR-compliant software components into a body electronics ECU of a series production vehicle. Now AUTOSAR is on the road, and Volkswagen and IAV use Model-Based Design with Embedded Coder™ for production.
This presentation covers our experiences with function development and integration of a Drive Select application into a gateway ECU of the Volkswagen MQB platform.
Dr. Sven Semmelrodt and Michael Kreutzer, Continental Automotive GmbH
In the field of automotive visualization electronics, such as instrument clusters or secondary display units, commercial vehicle OEMs are demanding increasingly sophisticated HMI designs with respect to graphics and functionality. In response, Continental Automotive provides gravis (graphic and video system), a development platform to simplify and speed up the design, development, and test of HMIs.
The gravis platform applied with Simulink helps to define the HMI requirements together with the OEM. To support code generation for embedded target hardware, the export interface generates TLC files for each gravis function block. Hence, Embedded Coder can be employed for generating code from the Simulink HMI design.
This session describes the capabilities of grADI (the HMI design tool); the export interface based on function libraries, S-Function, and TLC technology; and the benefits of using gravis incorporated into the MATLAB and Simulink environment.
Daniel Baumgärtner and Dr. Gregor Dasbach, Robert Bosch GmbH
The Bosch eBike system entered the market in spring 2011. Today it is considered a benchmark due to its drive performance and its excellent responsiveness. A growing number of bicycle brands offer e-bikes with the Bosch system. During the development process, we faced many challenges that are atypical for automotive standards, such as an extremely short time to market. To enable an agile development process for the drive controller, a specific tool chain was established, starting from model-based virtual prototyping based on Simulink, through dSPACE® AutoBox-equipped prototype bicycles and test benches, to code generation using Embedded Coder. Using this tool chain, we were able to design and optimize highly complex software modules and to reduce efforts required to make changes.
The presentation gives insight into the challenges of this fast-growing market segment of e-mobility and how agile software development is employed to meet the requirements of consumer product life cycles.
Dr. Riaz Ahmed, Carts GmbH
Permanent magnet synchronous motors (PMSMs) need magnetic flux position feedback for the torque or speed control algorithm. Conventionally, this magnetic flux position feedback system consists of N pairs of permanent magnets arranged in a ring shape and two Hall sensors. This position feedback system has many problems associated with it—including inaccurate magnetic pole lengths, inaccurate distribution of magnetic field strength along the magnetic rings, electromagnetic field influence on the Hall sensors, inaccurate axial position of two Hall sensors, error due to radial separation between two Hall sensors, and temperature and applied voltage variation between two Hall sensors—that produce inaccuracies in magnetic flux position calculation. The N pair of magnetic poles and ring structure make this setup very expensive.
This session presents a new setup consisting of a cam mounted on a rotor shaft, a pair of magnetic poles, a single 2-axis Hall sensor mounted on cam link rod, and software algorithms to calculate accurate magnetic flux position. The cam has grooves equal to half of the number of pole pairs. A one-way trip of the Hall sensor over the magnetic pole pair will determine the position of the rotor permanent magnetic field for one pole pair. This setup can be outside the motor casing, reducing electromagnetic influences on Hall sensor signals.
Quang Minh Tran and Dr. Ingo Kreuz, Daimler AG
In the world of textual programming, especially object-oriented programming, refactoring has become an established technique for improving the structural quality of code. With refactoring, the structure of a code body is improved through small restructuring steps without its external behavior being altered. Refactoring Simulink models by restructuring them is also a very common, yet highly time-consuming and error-prone activity. Unfortunately, tool support for refactoring Simulink models is limited. In this session, we present the idea and first results of our current effort to develop a refactoring technique with tool support for Simulink models.
Optional Technical Session
Patrick Munier and Stefan David, MathWorks
Functional safety standards such as IEC 61508 and ISO 26262 require dedicated activities related to code verification. Embedded software engineers, project managers, and quality assurance managers are involved in the process of matching objectives of the standards with operational constraints at the development process level. This session discusses Polyspace® code verification for achieving compliance with these IEC and ISO standards and for improving the quality of the embedded software.
Wednesday, 18 April, Customer Sessions
Manji Suzuki, DENSO CORPORATION, JAPAN
ISO 26262 demands strict work product management in configuration management, change management, and traceability management for automotive embedded systems, creating concern about drastic manpower and cost increases required to deal with ISO 26262.
At DENSO, we think that "process", "method," and "tools" are important to managing complexity and mitigate risks. Tools have especially become a key way to realize effective and strict management with minimum additional efforts.
DENSO started evaluating open source software and application life-cycle management (ALM) solutions based on Eclipse technology in 2003; however, we finally decided to adopt PTC (former MKS) Integrity as the corporate standard ALM tool not only for ISO 26262 but also for leveraging our capability.
This presentation shows DENSO's use of PTC Integrity for projects using legacy processes and provides experiences with applying it also to Model-Based Design.
Thomas Burger, Ralf Müller, and Marco Kunze, Continental Automotive GmbH
Nigel Tracey and Mark Rutter, ETAS GmbH
This session presents our experience using on-target bypass to efficiently validate and verify Simulink based System Design Automation (SDA) models. On-target bypass using Simulink models allows validation and verification of a new control strategy directly within the production ECU software. On-target bypass has previously not been feasible because it was difficult to integrate code for a new Simulink model directly into the production ECU software without the support of the ECU software team. The EHOOKS tool from ETAS now makes on-target bypass feasible by automatically building an updated ECU software version that includes the new Simulink model integrated in a defined and controllable manner.
On-target bypass, in combination with SDA supporting the powertrain architecture, offers an easy and powerful way to execute, test, and precalibrate the model-based functions directly on the ECU.
Vehicle Dynamics Control Systems: A Case History of In-House-Developed Strategies, from Concept to Ferrari GT Applications
Stefano Varisco, Ferrari
Ferrari GT approaches vehicle dynamics development as an opportunity for continuous performance improvement. One of the most effective approaches is to develop high-level strategies for vehicle dynamics control systems mainly in-house. During each phase of the concept development, code is checked to confirm it meets the strict requirements for robustness and reliability.
This session presents outcomes related to the most recent vehicles introduced into the market; the performance of GT cars has been significantly increased by fitting modern control techniques to existing control hardware.
The requirements for code generation and verification are addressed by a complete tool chain of software covering all the development phases. Following the spirit of ISO 26262, this process begins with Simulink models and its own validation, goes to automatic code generation and verification, and ends with the simulation of the target hardware.
The experience gained will be immediately applied to other real cases, with the goal of improving the overall performance and quality of our product.
Alexander Much and Thomas M. Galla, Elektrobit Automotive GmbH
The automotive standard AUTOSAR provides a standardized basis for ECU software development consisting of a layered software architecture with over 80 software modules and libraries accompanied by an associated development methodology.
The AUTOSAR software architecture clearly distinguishes between application software components (SWCs) and basic software (BSW) modules. The description of interfaces and internal behavior of software modules and components is contained in standardized AUTOSAR XML files.
Among other things, these description files contain information on the scheduling, concurrency of modules, the call graph, and the use of specific memory sections. This information can be used as the basis for the verification of the ECU software.
In this session, we present an approach for ensuring code is free from run-time errors for AUTOSAR-based ECUs and for using standardized AUTOSAR XML files to make the verification of software possible and computationally feasible.
Because more and more automotive ECUs are subject to safety standards such as IEC 61508 or ISO 26262, such verification is an important part in the documentation of the safety case.
Customizing Modeling Guideline Checks Within a Continuous Integration Framework for Automotive Model-Based Control Software Development
Dr.-Ing. Axel Schloßer and Dr.-Ing. Philipp Orth, FEV GmbH
Johannes Richenhagen, RWTH Aachen, Institute for Combustion Engines
Software development for automotive control units faces various challenges. From the market perspective, many product variants, short development cycles, and a high degree of innovation are required to flexibly react to customer needs. From the technological perspective, upcoming technologies such as variously electrified drivetrains, alternative fuels, and safety and comfort functions with powertrain intervention make the handling of control complexity necessary.
In reaction, development process requirements are formulated and standardized by reference process models such as Automotive SPICE, CMMI, or ISO 26262.
If software is applied to several projects for a longer period of time, the product quality degrades. For suitable countermeasures, an integrated quality assurance approach is necessary.
A testing strategy must involve adequate verification and validation methods and needs to be realized by an agile framework that allows for minimized manual testing effort, continuous software inspection, and traceability of test results over the entire project to facilitate frontloading of systematic software error mitigation.
In this session, we present a framework that aims to cover these demands.
Design of a Functional Size Measurement Tool for Real-Time Embedded Software Requirements Expressed Using a Simulink Model
Jori Le-Bihan and Boubker Bouzid, Renault
The use of ECU software in cars has grown considerably in recent years. To more tightly manage the development costs of software suppliers, as well as its own validation activities, Renault SAS measures the functional size of our ECU software specifications using international standards, and uses this information as the main input for estimating software development costs and schedules.
The COSMIC functional size measurement method is a standardized method for measuring the functional size of software. To obtain this size and reduce measurement differences caused by interpretations of individual measurers, we propose a functional size measurement (FSM) procedure for real-time embedded software with specifications modeled using Simulink. The design of this FSM procedure is based on the mapping of key concepts in both Simulink and COSMIC and on the identification of the mapping rules for extracting required information stored in the Simulink files. This procedure has been implemented as an automated measurement tool for software requirements specified using Simulink. This COSMIC functional size measurement tool has been applied at Renault SAS on real ECUs.
This session presents the design of the FSM procedure based on the COSMIC method to obtain the functional size of software specifications expressed with Simulink. This includes COSMIC rules tailored for Simulink models and the explanation of the mapping of the Simulink concepts to the COSMIC concepts. This session also presents the automated measurement tool developed at Renault SAS for implementing this approach and its application on real powertrain ECU software models.
Model-Based Development as a Key Element for Continuous Deployment Within Electric Propulsion Systems
Dr. Jonn Lantz, Volvo Cars Corp.
During the past decade, Volvo Cars Corp. has carried out numerous Model-Based Development (MBD) projects, all providing different experiences of the impact of MBD. Today, the present connectivity and electrification challenges call for a more time-efficient and software-oriented development strategy. The goal is continuous deployment, in which developing new features at R&D, deploying them on real or virtual cars, and getting real(istic) customer feedback to R&D can be completed within days.
For Volvo Cars, this means changing a large and complex organization often used to requirements engineering, with modeling tool chains still under development and, considering the electrification, lots of new technology. Nevertheless, ongoing projects with key groups at R&D in close collaboration with research teams at Göteborg University and Chalmers University aim to deliver proof of concept using seamless model-based integration and VR rigs. We also consider AUTOSAR, providing standardized software architecture and facilitated supplier collaboration, as a plausible road to large-scale seamless MBD. The present challenges include large-scale model-based integration, automation, unified modeling standards, and scalable agile MBD.
We gratefully acknowledge financial support from the Swedish innovation agency Vinnova and the XEVCO project.
Raymond Tinsel, DAF Trucks NV
Co Melissant, MonkeyProof Solutions BV
At the MathWorks Automotive Conference 2010, DAF Trucks and MonkeyProof Solutions presented the challenge DAF Trucks was facing with the increasing importance and complexity of software in vehicle development, the chosen strategy (migration to Model-Based Design), the transition and migration needs, a tool suite, and some early findings on the chosen approach and environment.
In this session, we will present our progress, tangible results, and findings two years down the road. Perspectives that will be covered are design quality and project management in vehicle development; the design environment and tool suite for design data management, traceability, and testing; and ISO 26262 reflections on the current workflow. We share results yielded by the adoption of Model-Based Design and discuss how project management is supported by a well-defined workflow and supporting infrastructure and design environment. We also share an evaluation of the workflow and design environment in light of ISO 26262.
Christian Corvino, Automobili Lamborghini SpA
This session shows how Lamborghini leverages the MathWorks tool-chain by integrating automatically generated C code into the parent standalone embedded project, implemented on the Freescale™ MPC55xx automotive microcontroller. The application is the Variable Valve Timing (VVT) control algorithm for the new Lamborghini Aventador engine control unit. After modeling the VVT system in Simulink using the energetic approach technique called Power Oriented Graphs and designing the controller, we used Fixed-Point Toolbox™ and Embedded Coder to generate the production C code. According to the V-diagram, some steps were completed before generating the code automatically. This approach saved time and cost within the validation process during the engine test bench. This challenging goal was achieved thanks to the simplicity of iterating the code generation process.
Each 15-minute session is led by a MathWorks domain expert and includes 5 minutes for Q&A.
- Parallel Computing with MATLAB and Simulink
Dr. Frank Graeber
- AUTOSAR for Production Projects
- Videos, Images, Signals? Stream Processing in MATLAB with System Objects
Dr. Frank Graeber
- Physical Modeling Introduction
- Advances in Applied Control Design
- Traceability to Requirements Docs Made Easy
Dr. Marc Segelken
- Simulink for Work Groups Using Simulink Projects
Dr. Joachim Schlosser
- ISO 26262 and Model-Based Design - A Done Deal?