MathWorks
AUTOMOTIVE Engineering CONFERENCE 2020

10:00 - 10:30

A shift from 'oil and gas' to 'energy' is taking the industry out of its comfort zone, but digitalization provides a way to manage transition risks. practical projects. Organizations have defined their high-level digital transformation objectives and are now looking to their engineers and scientists to achieve them. This will involve learning new technologies, collaborating with unfamiliar groups, and proposing new products and services. To meet this challenge, energy companies must master how to systematically use data and models, not only during the research and development stages, but also across groups throughout the lifecycle of the offering. An effective digital transformation plan needs to consider changes in people’s skills, processes, and technology.

12:30 - 1:00

Unconventionals, particularly shale resources, form a large and increasing part of our national energy reserves, requiring careful planning and optimization for efficient production. To this end, large quantities of data is being acquired about the field and all stages in the life of each well, including geological, drilling, completion, and production over the lifetime of the well. However, utilization of this data to optimize well design and performance poses a significant challenge, since much of the acquired data is at the surface, and hence only provides a noisy and indirect estimate of a deep (~2 miles) and long (2-3 miles) lateral well with 10s (20-50) of stages in an aggregated/averaged manner. Optimizing decisions on a per-stage basis, which is how the well is completed, requires estimating production from the well on a per-stage basis, which is neither measured for most wells, nor is easily inferred from the data acquired.

In this work, we demonstrate the use of clustering and optimization methods to construct predictive models of per-stage production using a collection of stage attributes derived from geological information, surface measurements during drilling and fracturing operations, along with per-well production data. Our analysis reveals the effectiveness of fracturing on a per-stage basis, thereby enabling better decision-making for future wells given the stage feature data acquired before production. The model incorporates methods to reduce/manage the effect of over-fitting while preserving accuracy, and also provides recommendations for experiment design and data-driven optimization of both planning and operational decisions. We show numerical results from the model applied to synthetic/simulation data. MATLAB’s capabilities for efficient analysis and visualization of large datasets proved very valuable in demonstrating our concepts.

Successful application of AI to engineering requires a complete design workflow rather than data and algorithms. This presentation will discuss how MathWorks developed a complete enterprise engineering platform for AI.  You will see the reasons for Gartner to name MathWorks a “Leader” in the 2020 Magic Quadrant for Data Science & Machine Learning Platforms.  

With the industry demanding new automotive technologies be delivered at an ever-increasing rate, Ford has embraced both MBD and MBSE to help deliver these technologies. With MBD as the foundation, it became essential for Ford to have well-defined MBD processes and tools in place. The Ford Motor Company MBD Core team was established to do just that. Through agile processes and lifecycle planning, the team develops and deploys common MBD processes and tools for all users globally at Ford. Our belief is that with a centralized MBD support team, we are not only developing common processes, best practices, and lessons learned for MBD, but we are also laying the groundwork so that we can effectively perform virtual vehicle level testing.

Massive change is underway in the automotive industry with trends in autonomous driving, vehicle electrification, and connectivity. In this talk, Ramamurthy Mani shares how MathWorks is addressing complexity, scale, and collaborative workflows in tune with evolving demands on automotive software architectures.

Service-oriented architecture (SOA) is a software architecture based on the concept that a system consists of a set of services in which one service may use another, and applications use one or more of the services based on their needs. SOA promotes a loosely coupled component-based approach using middleware for service-oriented communication. SOA concepts are used in multiple industry standards, including: AUTOSAR, ROS and DDS.

In this talk, MathWorks showcases how the Simulink^® is used to model and simulate application software based on SOA. The presentation highlights:

• Modeling of message-based communication between software components
• Modeling of Adaptive AUTOSAR software components
• C++ production code generation with Adaptive middleware interfaces (ara::com), and AUTOSAR XML export

A123 integrated their Model-Based Design toolset for AUTOSAR development with Jenkins for continuous integration. This was done to support the expanding team and usage of models. The resulting environment automates 90% of the steps along the common software development process, and also facilitates design reviews that are based on a rich set of data and metrics.

MathWorks has helped many teams migrate their software development processes to meet ISO 26262. In the process, we also helped these teams optimize the use of Model-Based Design methods and tools. We identified a few areas of common pitfalls when executing the migration, such as formal process mapping, artifact generation, software architecture design and review, and tool qualification. We would like to share our learning and recommended solutions through this presentation.

Navistar developed a new Data Dictionary tool (DD) enhancing Simulink’s Data Dictionary to simplify control model design while supporting production code generation and deployment. Providing further data object abstraction, the DD simplifies various user groups’ processes and streamlines collaboration during development. For example, function developers use the DD to specify functional requirements for data objects such as names, physical units, and signal value ranges; while embedded software engineers simultaneously specify production code specific aspects. This approach lets function developers focus on functional requirements while embedded software engineers focus on production code implementation and integration. The DD also includes a lightweight component-based framework supporting the integration of hundreds of software components for both AUTOSAR and non-AUTOSAR applications using a bottom-up approach.

Many industries are embracing Big data – the collection and accumulation of large amounts of information. Automotive, aerospace and large equipment manufacturers are seeking to understand customer usage and profile equipment fleets data. The data will help in the design, engineering and production of solutions customers will find desirable and compelling. The solution will be about how to use such large data at scale with the utilization of a Spark Computation environment in conjunction with the interface MathWorks is developing.

In the past, engineers download terabyte-sized ADAS datasets to look for edge cases. This approach consumes huge amount of network bandwidth and local storage space. We created a new and more efficient way, which utilizes MATLAB to access Apache Spark resources to decode, analyze data, and search for edge cases right on the Hadoop file system. It dramatically improves throughput and reduces the amount of data downloaded to the engineer’s workstation.

This approach was successfully used to analyze ADAS feature usage from the CAN traffic on Ford’s Big-Data-Drive fleet of vehicles. It will be deployed for all future Big-Data-Drive vehicle analysis.

Do you have a strategy to analyze the data from your connected test vehicle fleet? How fast are you able to develop and apply analytics on huge sets of data to find desired events or find trends that were previously unknown? Are you able to work with all of your data instead of a subset?

In this talk, Will Wilson will demonstrate how to implement a workflow with MATLAB^® that addresses these issues. Topics include:

• Exploring the types of questions you can ask of your fleet data
• Preparing your data for efficient analytics
• Developing analytics that execute on a “per unit” or “across all” basis
• Deploying analytics to keep up with the continuous intake of test data

The rapid boom in big data generation, data science, and machine learning has led to a massive opportunity for continuous improvement, optimization, and automation across virtually all industries. The focus here is on applications of machine learning tools within the steel industry, demonstrating the capabilities, speed, and accuracy using MathWorks^® Deep Learning Toolbox. Two steel specific case studies are presented: automated non-metallic inclusion classification and coarse dimensional measurement of in-process steel production. The technique of transfer learning was employed to reduce computational overhead, and it was found that a proper ground truth training dataset with intelligent image pre-processing yielded results with better-than-human accuracies at vastly superior speeds. Implementation of finished models in a standardized dynamic-link library format provided seamless integration with other common programming languages, which led to a straightforward, easily scaling, production roll out. Advantages were immediately apparent with regard to task specific man-hours and evaluation consistency. Basic architectures, pre-processing steps, training parameters, and model performance are described for each case study.

Machine learning is a hot topic in the automotive industry. Deploying machine learning algorithms to electronic control units (ECUs) is often a bottleneck because of the memory, CPU throughput, and software development and integration techniques required to support machine learning algorithms.

In this presentation, Gokhan Atinc provides an overview of machine learning technologies and deployment workflows for embedded processors. He will also discuss advanced capabilities that are of interest for automotive and adjacent industries, including in-place modification support, Simulink^® support, and fixed-point conversion.

MATLAB^® has scaled up to support cluster and cloud-based data and computing frameworks. In the meantime, data and computing framework technologies continue to evolve rapidly. In this presentation, Arvind provides an example of how an enterprise customer integrated MATLAB with their existing framework. The integration enables engineers to slice and dice very large datasets using Apache Spark™ and extract forensic slices to develop analytics that can then be pushed down to execute at scale on the cluster. The integration with MATLAB also supports workflows that conform to enterprise-level security, governance, and access controls requirements while enabling users to make the results of their analytics easily accessible across the organization.

With the advancement in machine learning, access to data with V2X connectivity, and more reliable plant model simulation, reinforcement learning has been considered recently as a control design option for the feedback control of automotive systems. In this talk, the challenges of applying classic control methods with focus on PID structure are briefly discussed and a perspective to deploy reinforcement learning to address some of these challenges is presented.

In this presentation, you will learn how to easily create 3D scenes and road networks for automated driving simulation using RoadRunner. You will also see how to automate the creation of road networks from high-definite (HD) maps.  After you finish creating and editing the scenes, you can export them to industry standard file formats such as OpenDRIVE and automated driving simulators such as CARLA, NVIDIA DriveSIM, and Simulink.