MathWorks Automotive Conference 2014

Model-Based Design in a Seamless Embedded Software Process

8:55–9:25 a.m.

Model-Based Design and automatic code generation are considered to be the standard for embedded software across many industries. Much focus is given to the control system design and simulation tools, but these are only part of the software development process. To be successful in Model-Based Design, the whole toolset and development process must work together to facilitate success for every engineer. Developments over the last 10 years have transformed the ability of Model-Based Design to support production applications with an efficient implementation and predictable process. In this keynote presentation, Craig explores progress made and areas for continued development.

Craig Stephens, Ford Motor Company

Implementation of ISO 26262: Adoption, Challenges, and Efficient Application

9:45–10:10 a.m.

The international standard ISO 26262 was created as the core of functional safety guidance in the automotive industry. Since its publication in November 2011, many OEMs and suppliers around the world have adopted it and accumulated many important experiences. This presentation shares our observations on the obstacles and challenges companies typically struggle with as they consider functional safety and adopt ISO 26262. We also share our view on the technical and organizational competencies companies should have in place in order to develop products and systems successfully under the standard, particularly in light of their roles in the supply chain (OEM, tier I, tier II, component and/or tool supplier, etc.). Lastly and most importantly, we look at the impact that development tools have on the successful design and release of automotive systems under the standard.

Bonifaz Maag and Mike Staszel, Kugler Maag

Analytics Activities at Ford Motor Company

10:40–11:05 a.m.

Incorporating rigorous analytics, such as machine learning, operations research, data mining, and big data, throughout the business has played a key role in the resurgence of Ford Motor Company in recent years. Analytics are used widely in diverse applications at Ford, including research, product development, manufacturing, supply chain, marketing and sales, finance, purchasing, information technology, and human resources. Ford received the 2013 INFORMS Prize in recognition of its long-running company-wide effort to use data science and predictive analytics to improve overall operations and performance. In this presentation, Yu-Ning introduces current analytics activities at Ford featuring examples of successful analytics applications.

Yu-Ning Liu, Ford Motor Company

Validation of Aftertreatment Temperature Requirements Using MathWorks Tools

11:25–11:50 a.m.

Engine aftertreatment temperature is critical for proper operation to meet increasingly stringent emission regulations. Acceptable aftertreatment temperatures and regeneration operation need to be verified for the numerous applications seen in heavy-duty construction, industrial, and mining applications. The development of models using MATLAB® and Simulink® has allowed significant cost reductions by eliminating cycle tests in test cell and replacing them with simulation. The robustness of the system was evaluated using validated models to assure proper system performance with expected component variability.

Travis Barnes, Caterpillar Inc.

Design, Analyze, and Implement Radar Sensors’ Alignment Algorithm with MATLAB

1:00–1:25 p.m.

Radar sensors can detect a target’s range, range rate, and azimuth in the vehicle coordinate system (VCS) to enable adaptive cruise control, forward collision warning, and other features. To obtain accurate target properties, a radar’s boresight should match the designed pointing angle in the VCS; misalignment of the boresight can become result in errors on azimuth that must be compensated to achieve accurate target lateral offset in the VCS. Delphi designed a least-squares solver in MATLAB to compute misalignment angles and analyze vehicle testing logs, and generated production code with MATLAB Coder™.

Liang Ma, Delphi

Enabling Model-Based Design: Robust Collaborative Development of Embedded Systems

2:35–3:00 p.m.

Model-Based Design and automatic code generation have enabled a dramatic reduction in the defects, effort, and cost of software development for embedded systems. This is especially true for individual control functions, which by nature have a limited scope. However, the robust collaborative development of embedded control systems continues to present serious challenges to the realization of truly ubiquitous Model-Based Design. This talk provides use case scenarios to illustrate how these challenges can be overcome through process discipline; centralized version-controlled data and architecture management; automated integration and consistency checking; and object dependency visualization.

Bill Milam, Ford Motor Company

Model-Based Design Maturity: Benchmarking the Automotive Industry

9:25–9:45 a.m.

Model-Based Design is widely adopted for complex system and embedded software development. However, there is a wide range of Model-Based Design maturity levels among engineering organizations. As a result, some organizations achieve greater development efficiency and product quality gains than their peers. Drawing on data from Model-Based Design Process Assessments™ and survey results, this presentation describes methodologies; outlines characteristics of leaders, average adopters, and laggards; and closes with a look at trends. The presentation also reviews:

  • Types of assessed automotive OEMs and suppliers
  • Key characteristics of leaders, what they do that makes them leaders, benefits they gain, and trends
  • Comparison of the automotive and aerospace industries’ maturity

Vinod Reddy, MathWorks

From Big Engineering Data to Insights Using MATLAB Analytics

11:05–11:25 a.m.

MATLAB provides a strong platform for the analysis of complex engineering data. This presentation describes how to use MATLAB analytics to enable deeper insight into the logged data for vehicle design optimization, prognostics, diagnostics, and other applications. These analytics often need to operate on big data sets from different sources, vendor technologies, and file formats. A few typical use cases are discussed along with reference architectures that enable scalability of analysis with big engineering datasets.

Arvind Hosagrahara, MathWorks

Eliminating Design Errors in Your Algorithm Using Simulink Design Verifier

1:25–1:45 p.m.

Many engineers test their algorithm models in simulation. Through simulation they identify design and requirement errors in their model before generating production code. However, eliminating design errors remains a challenge and extensive testing with 100 percent coverage may still result in a design that contains robustness errors such as overflows and divide-by-zero. Some of these errors may reveal themselves under rare conditions and could be time consuming to debug. They may be induced by certain calibration values and only be found on the HIL bench or in a test vehicle. This presentation illustrates a method for detecting and eliminating such design errors using Simulink Design Verifier™.

Nishaat Vasi, MathWorks

Modeling HDL Components for FPGAs in Control Applications

1:45–2:05 p.m.

FPGA is an emerging technology for automotive control applications to implement peripheral and algorithmic components requiring fast turn-around time. In this session, learn techniques to model HDL components that can be implemented on an FPGA. A motor control example is used to demonstrate common tasks such as:

  • Designing a new component and generating HDL code
  • Integrating components that will be executed at disparate rates
  • Migrating a component from a C to an HDL specification

The components used in this example were modeled in Simulink, tested with simulation, and verified on real-time hardware.

Mark Corless, MathWorks

Accelerating Optimization, Test, and Code Generation with Parallel Computing Toolbox

3:00–3:20 p.m.

Development of system and controls involves extensive use of simulation for design exploration, optimization, and verification. Increasing design complexity has led to challenging computational demands. In many cases, engineering tasks can be automated, parallelized, and distributed to take advantage of latest computer hardware and reduce execution time. This session introduces Parallel Computing Toolbox™ and MATLAB Distributed Computing Server™, including exemplary workflows that illustrate how to accelerate code generation, design optimization, and test automation with Simulink.

Ryan Chladny, MathWorks

Real-Time Simulation of Battery Packs Using Multicore Computers

3:20–3:40 p.m.

This presentation focuses on a technique to scale battery models from cell-level to pack-level and the subsequent preparation of the battery pack model for real-time simulation. A MATLAB function creates a battery pack model of arbitrary size and connectivity. The function partitions the model for concurrent execution on multicore real-time computers. The flexibility of scripting the creation of pack models allows the user to efficiently test multiple configurations for optimal utilization of multicore targets, including load balance, data transfer latencies, and scheduler overhead.

Javier Gazzarri, MathWorks

Developing AUTOSAR Compliant Embedded Software

2:35–3:40 p.m.

This master class provides an introduction to AUTOSAR compliant embedded software development. After a short overview of the AUTOSAR standard and its implications for development tools and processes, the workshop focuses on mapping Simulink constructs to AUTOSAR definitions, performing model verification, and configuration of Embedded Coder® with a goal of interfacing with AUTOSAR authoring and RTE generation tools.

Data Analysis Using MATLAB

3:50–4:55 p.m.

This master class shows you techniques for using MATLAB to analyze and visualize data, as well as share results. You will see how to:

  • Access data from multiple files
  • Use interactive tools for iterative exploration, design, and problem solving
  • Automate and capture your work in easy-to-write scripts and programs
  • Generate reports automatically for documenting, presenting, and sharing your results
  • Develop a graphical application for analyzing multiple data files

Sean de Wolski, MathWorks

AC Motor Control Architecture, Code Generation, and Verification

3:50–4:55 p.m.

This master class uses a field-oriented control algorithm for a permanent magnet synchronous motor to illustrate how to generate efficient fixed-point C code from the controller model, integrate it with handwritten code for the embedded device drivers, and use the fully integrated software to spin the motor hardware. Topics include model architecture, algorithm export, scheduling techniques, code profiling, and code verification using processor-in-the-loop (PIL) testing. The data dictionary capability in R2014a is also covered.

Jeff Tackett, MathWorks