Keynote: Higher Efficiency with Scalability in Semiconductor and Mixed EV Architecture

9:10–9:35

We are in the transition into clean, safe, and software-defined mobility, with new high-voltage power technologies making electrified vehicles more competitive and governments pushing for CO2 neutrality. This is a disruptive paradigm shift for established OEMs. The next challenge is to deliver the promise of software-defined vehicles, which requires increasing computing power, managing the data complexity within the car, and connecting the car with IoT/clouds. 80 percent of this innovation is based on semiconductors that power the next generation of vehicle electrical/electronic architectures.

These new architectures need to address real-time, functional safety and security while distributing and computing gigabytes of data from the sensors across the car backbone to enable innovative applications such as autonomous driving. Additional requirements include safe power distribution, redundancy, and diagnosis to guarantee “Always Power On” for operating the car. Lastly, software-defined vehicles will need to provide customers with new services that will be constantly updated over the air. The semiconductor industry is already providing a variety of products, including SoC and MCU, which address these challenging requirements and support mixed EV architectures.

This hardware offering needs to be complemented with modern R&D processes in which simulation and modern SW update management are standard elements. Infineon is committed to push digital twins and provide models to an environment which allows faster development, enabling design verification and approval processes at an early stage. With this, Infineon can provide a secure supply and shorter time to market and deliver the latest and most appealing technologies that the next generation of customers is expecting.

Hans Adlkofer

Hans Adlkofer, Infineon


Keynote: Accelerating Development of Clean, Safe, Automated Software-Defined Vehicles 

9:35–10:00

In this talk, Andy Grace, the VP of Engineering at MathWorks, will share his vision on how Model-Based Design is evolving to accelerate the design and development of clean, safe, autonomous software-defined vehicles. Through this talk, the audience will learn about the impact that MathWorks is having in the automotive industry and the investments that are in the making to address new industry challenges. 

Andy Grace

Dr. Andy Grace,
MathWorks


SDV: Integrating Simulink C++ Generated Code in Android Automotive Environment

10:00–10:25

High-performance computing and service-oriented architecture (SOA) are emerging as the technological foundations to provide the computing power and the high level of abstraction required to accelerate the development of software-defined vehicles. Abstraction from the hardware is typically provided through middleware and operating systems such as Android Automotive OS. In this context, the integration of application software with the middleware is crucial to establish an agile, automotive-grade software development process.

This presentation from Renault demonstrates the ability of Simulink® to generate C++ application code that can be integrated and run in Android Automotive OS environment using Android Binders and the Android Interface Definition Language.

Based on SOA concepts, this demonstration shows a control loop model that uses methods and events for service-oriented communication.

Remy Brugnon

Rémy Brugnon, Renault Group


Software Architectures and Virtual Integration in Model-Based Design

11:00–11:25

The rapid increase in automotive system and software complexity demands a design process that can help maintain consistency and traceability all the way from high-level system design to software implementation. Model-Based Design approaches offer distinct advantages that can help go from software architecture conceptualization to deep component design and back up to software integration and testing, enabling higher quality and reliability of the final product. In this presentation, you’ll see how Simulink® is rapidly evolving to realize this vision. Recent investments have helped create an environment that enables the detailed design and simulation of software architectures and easy integration of models and code components for early virtual integration.

Ramamurthy Mani

Ramamurthy Mani, MathWorks


CI/CD 2.0: From Scripted Jenkins Pipelines to Process Advisor

11:25–11:50

At Continental, we implemented the first Jenkins-based continuous integration pipeline in 2017 to accelerate agile, model-based vehicle software development. That pipeline was then heavily customized over the years. As a result of this customization, we had to spend more and more maintenance effort for CI pipeline itself, making any process update very time-consuming.

We decided to use the new Process Advisor App to rebuild and simplify the Jenkins Pipeline and to bring all Jenkins-specific parts of the pipeline into MATLAB®, including JFrog Artifactory and Conan.

In this presentation, we demonstrate the different aspects, findings, and advantages of this new approach.

Martin Römpert

Martin Römpert, Continental Automotive


VDA SIL Standard: Change in SW and System Development in the Automotive Industry

11:50–12:15

Software-in-the-loop (SIL) is the enabler of continuous and agile development processes in the automotive industry. Without SIL, the release of open-context systems with a high number of combinations of possible input data (such as automated driving) is practically impossible. In the future, it can be assumed that most software tests in the automotive industry will be carried out in SIL environments. To work across companies, domains, and platforms, it is necessary to standardize the SIL interfaces across the industry. Without standardization, many individual interfaces would be created. The effort and costs involved would be enormous. SIL standardization is an essential requirement for successful virtualization and automation of software testing in the automotive industry. The VDA project group SIL Standardization has been publishing standards for technical interfaces of the SIL infrastructure since the end of 2019.

MathWorks supports the standardization effort by actively participating in the VDA project group and will give an overview of upcoming features for FMI 3.0 as a base for Simulink® as an integration platform.

Amir Sardari

Amir Sardari,
Bosch

Gernot Schraberger

Gernot Schraberger,
MathWorks


ChatGPT and Large Language Models with MATLAB

12:15–12:30

Learn how large language models (LLMs) work and build a transformer model in MATLAB. See a demo of an LLM-based model for MATLAB and how you can use it in your work, including which prompts to use.

Deborah Ferreira

Deborah Ferreira, MathWorks


Women in Tech Forum: Lunch and Networking

13:30–14:00

As part of the Women in Tech initiative, MathWorks will be organizing a meeting point within the exhibition area at this year’s MathWorks Automotive Conference. Interested delegates can learn about the initiative and DEI (diversity, equity, and inclusion) at MathWorks. Use this opportunity to talk to our technical experts, discuss experiences, and network with other industry peers.

Eva Pelster

Eva Pelster, MathWorks

How to Develop Model-Based AI Software for AURIX TC4x in MATLAB and Simulink

14:00–14:25

The traditional approach to embedded software development is burdened with many potential pitfalls like vulnerability to errors and time-consuming development due to manual peripherals configuration. This presentation introduces an alternative approach for embedded software development based on Simulink® models to overcome these disadvantages. Based on the automotive-driven use case of trajectory control, we want to give a best practice example for end-to-end development and deployment of AI-enhanced embedded applications. Using the Embedded Coder® Support Package for Infineon AURIX™ Microcontrollers, all necessary hardware components such as peripherals and memory units can be easily configured and simulated from the level of GUI. Learn how Simulink enables you to generate an optimized code for the target platform Infineon AURIX™ TC4x. The AI component will be compiled with a separate toolchain and combined with binary compiled out of generated code.

Mateusz Chmurski

Mateusz Chmurski,
Infineon


Electrothermal Modeling and Analysis of Battery Packs

14:25–14:50

The electric powertrain is becoming the leading solution in the passenger car sector. However, the components of the electric powertrain pose new development challenges for the automotive engineers. One of these components is the battery system. For its sizing, engineers need to consider different aspects such as the electrical behavior of the battery and the monitoring and management of its temperature. These challenges call for complex models that span different physical domains.

This talk demonstrates how engineers can model battery cells based on predefined or custom components. First, learn how to model a cell and—in the case of a custom cell—how to refine its electrical and thermal design as desired. The resulting cell model is then scaled up to build modules and complete battery packs with Simscape Battery™. For this scope, it is possible to use either an API or an intuitive interactive app.

In the second step, the resulting battery model is tested for performance and electrical and thermal behavior using selected test cycles. The model is then used to analyze the thermal behavior of the battery and design a liquid cooling system.


Powertrain Simulation for Concepts Evaluation, Range Estimation, and Calibration

14:50–15:15

The transition to zero emission and the demand for increasing levels of autonomy present many challenges for the development of modern heavy-duty vehicles (HDVs). Simulation plays a key role in accelerating the vehicle design stages, anticipating or preventing integration issues, and gaining competitive advantages.

In collaboration with MathWorks, MAN Truck & Bus has developed FASIMAN, a modular simulation tool based on Simulink for longitudinal dynamics, powertrain concept evaluation, energy consumption and range estimation. It uses energy management strategies and virtual testing of ECU functionalities and drivetrain calibration to support ADAS development and emission activities. FASIMAN’s modular approach enables it to model propulsion and vehicle configurations that are relevant for HDVs, including ICE, BEV, and FC.

In this presentation, see an overview of FASIMAN’s modular architecture and main features, including distinction between physical models and controls, bus system modeling, ECU integration (MIL), database connectivity, big data usage for load cycle derivation, and cosimulation.

We’ll also briefly present a fuel cell truck simulation for evaluation and design of the drivetrain concept, cooling concept, and supporting the control function development.

Christian Haupt

Dr. Christian Haupt,
MAN Truck & Bus


Streamline Automotive SPICE Compliance Using Model-Based Design

15:15–15:40

Automotive SPICE allows organizations across the automotive supply chain to assess and improve the capability levels of their own processes as well as those of their suppliers. Consequently, ASPICE-compliant processes allow suppliers to satisfy and even exceed customer expectations.

To effectively achieve ASPICE compliance, organizations choose to use the Simulink® product family. This is because automation capabilities in model-based systems engineering (MBSE) allow engineers to focus on their state-of-the-art products and innovations and leverage tooling support to achieve process quality aspects like traceability, consistency, and documentation.

Starting in R2022a, the IEC Certification Kit provides a mapping document between base practices of engineering processes in Automotive SPICE and use cases for Simulink.

In this presentation, see how these products work together in a streamlined way to support your ASPICE compliance efforts when it comes to system and software engineering processes.

Topics include achieving ASPICE compliance with MBSE by performing base practices of system engineering processes, maintaining consistency and traceability across your system architecture design and V&V artifacts, ensuring continuity of your system and software development artifacts, shifting your verification to the left and detecting issues as soon as they are introduced throughout the development process, and showing evidence for compliance with automatically generated artifacts.


How to Achieve Full Coverage of Configurable Code with Polyspace

16:10–16:35

Software is becoming more and more important in the semiconductor industry to further differentiate between vendors. Silicon companies are developing software to provide a software stack to leverage their silicon capabilities and to enable their customers to quickly jump-start their designs by providing a basis for ready-to-use software and middleware. Using software variants is becoming a popular method of supporting the diversity of possible hardware configurations, preventing source code duplication, and minimizing the footprint of the executable firmware.

Due to the heavy use of software variants, the exponential increase in possible software configurations is reaching the limits of traditional verification and validation methods. This presentation by STMicroelectronics proposes a novel framework that uses Polyspace® to analyze a selected subset of all software variants while providing the same guarantees as if all combinations had been analyzed, thereby reducing verification efforts without losing quality. This framework leverages results from structural code coverage to select the subset of software variants on which tests are executed and static code analysis performed.

Before releasing the software, developers can detect bugs and violations of coding guidelines for every possible software configuration, which would be impossible with either static analysis or dynamic test in isolation.

Cinzia Tomasello

Cinzia Tomasello,
STMicroelectronics


Agile Behavior-Driven and Test-Driven Development with Model-Based Design

16:35–17:00

Model-Based Design and the agile practices of behavior-driven development and test-driven development play an important role in modern, software-intensive, large-scale development projects. This presentation illustrates the benefits of a combined approach based on best practices established through years of work with engineering organizations.

Marc Segelken

Dr. Marc Segelken, MathWorks


Making the Most of FPGAs for Automotive Power Electronics Development

17:00–17:25

Designing efficient power electronics modules for EVs involves making informed choices about circuit topologies and semiconductor technologies and integrating them with sophisticated control algorithms that require low latency and high sampling rates. FPGA-based platforms have emerged as the preferred option for accelerating the development and testing of power electronics, allowing for real-time simulation while offering an attractive deployment target for control algorithms, both in the concept phase and in production.

Early collaboration between diverse teams is critical to the success of this type of project. In this talk, we discuss an integrated workflow that enables target-independent FPGA deployment of customized power electronics and e-motor plant models for real-time simulation with varying degrees of fidelity, including average-value, averaged-switch, piecewise-linear, and nonlinear models. Control algorithms can also be deployed automatically for rapid control prototyping. This workflow enables systems, algorithms, hardware, and software engineers to work simultaneously and efficiently, resulting in quicker iterations and fewer errors in system specification and implementation.

The final hardware target for algorithms is often unknown during the early stages of a project. To this end, this workflow allows for model-based hardware-software co-design or a seamless transition between hardware and software implementations. The implementation models can also be easily repurposed for different FPGA target devices or transitioned to ASIC since target-independent RTL code is generated. Finally, the toolchain is certified by TÜV SÜD to be suitable for ISO 26262, ASIL A–D. .

Dimitri Hamidi

Dimitri Hamidi, MathWorks

Virtual World Generation for BMW Driving Simulation

14:00–14:25

BMW operates the world’s most sophisticated facility for simulating real-life driving conditions with more than 14 simulators and usability labs aiming to provide the ideal simulation for every phase of the vehicle development process. Discover how BMW uses RoadRunner as part of their virtual world generation process to meet the requirements for high-fidelity simulation of real-life roads and traffic scenarios.

Hubert Cao

Hubert Cao,
BMW


Advanced Scene and Scenario Creation Workflows for Virtual Testing

14:25–14:50

Real-world scenario-based testing plays a critical role in ensuring the safety of ADAS/AD functions. Virtual testing is scalable in the cloud, efficient, safe, and repeatable. It enables you to test every incremental change of your function.

This presentation elaborates on some advanced capabilities of the RoadRunner product family in combination with Automated Driving Toolbox™.

The Scenario Builder for Automated Driving Toolbox support package enables users to quickly create scenarios from existing vehicle sensor data such as existing test drives or vehicle traces.

Once a seed scenario is defined, variations can be generated using the Scenario Variant Generator for Automated Driving Toolbox support package for SOTIF-related activities.

Engineers can bring in their own custom HD maps via the RRHD interface to build scenes.

Simone Hämmerle

Simone Hämmerle, MathWorks

Advait Valluri

Advait Valluri,
MathWorks


Deploying AI for Mission Profile Classification of Construction Equipment

14:50–15:15

A known challenge in designing construction equipment is identifying the correct mission profile while the vehicle is operating in the field. A mission constitutes a cyclical execution of a sequence of vehicle movements. Different missions can include similar movements with different execution speeds or arm positioning.

The main purpose of this project is to develop a model based on artificial intelligence that can recognize one of 20 possible vehicle missions of a wheel loader, directly on-board and in real time, thanks to the data collected from the available sensors.

Mission profile recognition is fundamental to enabling innovative applications such as:

  • Real-time, on-board controls for an optimized customer experience based on the current mission
  • Reliability analysis through failure contextualization
  • Market segmentation through understanding of customers’ behaviors

The identified mission can be used to automatically adjust vehicle parameters at runtime. For example, a pick-and-place maneuver can be supported by smooth aggressiveness for hydraulics, while hauling or stock piling operations can be supported by high aggressiveness. Alternatively, the vehicle control unit can suggest the best vehicle setting to support the current task through a message on the dashboard display. Furthermore, the identified mission profile is transmitted to a remote server to enable fault contextualization and to improve fleet management and maintenance services. At the server side, the faults are then correlated to the occurrence of certain missions to better identify the critical ones. This data is used to guide the vehicle design to better address such critical missions, significantly improving energy efficiency and vehicle reliability.

This project was developed by CNH Industrial in partnership with MathWorks Consulting Services relaying their expertise to design an AI solution with Statistics and Machine Learning Toolbox and using automatic code generation to quickly deploy the vehicle software.

Andrea Gravili

Andrea Gravili,
CNH Industrial


Demystifying DevOps: A Cloud Workflow for Fleet Analytics with Machine Learning

15:15–15:40

DevOps can sound imposing, but it can bring tremendous advantages to engineering workflows in the cloud by streamlining processes, adding maintainability and fostering a collaborative culture with IT and operations.

In this session, see a DevOps-style workflow for monitoring the state-of-health (SOH) of a battery fleet. Predictive maintenance—a powerful method built on machine learning techniques—can help to tackle this challenge, but it might be difficult or expensive to deploy and update the algorithms in the vehicle. We will demonstrate a cloud-based predictive maintenance system for monitoring the fleet that leverages streaming data to integrate your analytics with industry-standard technologies.

Learn how to develop an SOH prediction model and a drift detection model in MATLAB®. Using the Streaming Data Framework for MATLAB Production Server™, we can move from developing our algorithms with test data in file-based timetables to integrating with cloud streaming messaging services like Apache Kafka. Build and test these algorithms automatically using a CI/CD pipeline, deploying to an Azure Kubernetes Service cluster. Finally, the deployed system includes model performance monitoring using a dashboarding system for alerts and trends, and a time-series database to record historic data, which you can use to retrain your models. Using Infrastructure as Code techniques, we can easily update, rebuild, or roll back changes to the operational system.

See how engineering teams can thrive in a DevOps culture using MATLAB to operationalize their algorithms and bridge the gap between IT and Ops teams.

Nicole Bonfatti

Nicole Bonfatti, MathWorks

Martin Büchel Gravili

Martin Büchel,
MathWorks


Quality Inspection Based on Deep Learning and a Data-Centric Approach

16:10–16:35

Automated quality inspection of components in industrial production environments is one of the main requirements for achieving current productivity and quality goals. Since conventional rule-based inspection systems only partially meet these requirements and industry research projects in this area are hardly practicable, Miba AG has developed its own framework for the quality inspection of its high-quality components to achieve these goals in its production facilities. In this technical presentation, see and discuss the results of this framework based on deep learning with MATLAB®.


Service-Oriented Arbitration of ADAS Features with Model-Based Design

16:35–17:00

The main challenge of AD/ADAS is to imitate and outperform a human driver’s responsiveness to dynamic situations. The arbitration module in an AD/ADAS system plays a major role in making decisions for vehicle control.

The arbitration module is a central processing module that makes decisions based on track estimation, crash mitigation, and safety to orchestrate priorities and actuations of longitudinal, lateral, or driver control. This makes it a critical module. Any performance issue in this module could have a major impact on the vehicle’s safety and performance.

In signal-based architectures, an easy update of the application is not possible because it will impact the arbitration control scheme, which would lead to an entire software update and revalidation that’s not always possible or desirable.

By contrast, service-oriented architectures (SOAs) enable an easier update of an application or porting of a new application. The service registry and service discovery enable the client to locate new service providers easily, facilitating the integration of new features in the architecture.

Moreover, the arbitration in SOA improves the performance and quality thanks to high-speed information exchange among control applications.

This presentation includes a use case to compare the signal-based and service-based arbitration performances.

Darshana Unnikrishnan

Darshana Unnikrishnan,
KPIT Technologies

Nandakumar Kaiprath

Nandakumar Kaiprath, KPIT Technologies


Targeting GPUs for Automotive Applications

17:00–17:25

Many of today’s signal processing, image processing, and deep learning applications in the automotive field can benefit massively from using GPUs. This is equally true for production code and prototyping purposes. See how automated CUDA code generation from MATLAB® algorithms and Simulink® models can be used to accelerate applications such as lidar, camera, and radar sensor data processing. A complementary toolchain allowing MIL, SIL, and PIL simulations as well as performance analyses is key to efficient development workflows. Using the same Simulink model or MATLAB function, you can target both desktop GPUs as well as SoCs like the NVIDIA DRIVE® and NVIDIA Jetson™ platforms.