Math-Enabled Automotive Innovation
Tools for mathematical models and analysis are rapidly growing in capability and scope, enabling significant innovations in the automobile industry. These tools accelerate the development and implementation of new technologies into the automobile. The presentation reviews a number of recent innovations at General Motors that relied on new models or analyses, including wireless phone charging, the Global Human Body Model, Chevrolet Volt battery tray design, and ONSTAR Proactive Alerts. The models used and key results for each example are shared. Finally, Paul discusses opportunities for math to enable the future of mobility.
MathWorks Vision for Systematic Verification and Validation
Ad hoc simulation is an instrumental first step to gain an understanding of a system under typical operating scenarios, yet it has become inadequate for verifying designs of increasing size and complexity. Simulink® verification products extend and complement simulation to provide additional rigor, automation, and insight that your designs are functionally correct, comply with standards, and are faithfully implemented on target hardware. This talk explains the vision and expanding capabilities of these tools for dynamic testing and formal methods–based static analysis. Bill also discusses how to apply these techniques systematically throughout a production development process to achieve higher quality and productivity.
Driver assistance technology features require coordination and control of ECUs, sensors, and actuators, that are distributed functionally throughout the vehicle and in constant communication through CAN networks. The systems engineer is challenged to design, validate, and verify these distributed control systems, which are composed of hardware and software from disparate sources. Engineering specifications typically describe system logic requirements with natural language, tables, diagrams, and flow charts. These “non-executable” requirements allow for ambiguities, typos, and logical errors that introduce defects into software implementations. In this session, Nate demonstrates how creating executable models of system requirements in Simulink® has reduced software defects and improved product development timing. He also explores some best practices and tools developed for efficiently implementing the requirement modeling methodology.
The traditional development approach for a proof of concept (PoC) is to write native code for target embedded systems. It usually involves writing code in C/C++, cross-compiling for the target, downloading the code, and testing it in several iterations. The process could be long and tedious, and often constrained with limited hardware and software resources available on the target system. Alpine decided to try a new approach by using MATLAB® toolboxes in combination with MATLAB Coder™ to avoid such constraints and speed up PoC development in the early stages for one project. This presentation shares their experience, findings, and conclusions.
MATLAB® has proven to be an invaluable tool for system performance engineers dealing with today’s complex diesel systems and variety of applications. From calibration development to understanding field test and end-user performance trends, MATLAB is widely used within Cummins. This presentation discusses the variety of ways Cummins uses MATLAB for system performance integration and system performance analysis. This approach removes the roadblocks of complexity and leads to world-class delivery of diverse products to customers.
11:35 a.m.–12:00 p.m.
The success of a version upgrade depends upon the strategy used to implement it. The bulk of this presentation provides a recommended process for a systematic approach to enterprise-wide upgrades, designed to maximize access to benefits and minimize the costs, risks, and disruptions associated with achieving those benefits. These recommendations are based on practical experience in guiding enterprise customers in their upgrades to new versions of MATLAB® and Simulink®. In addition, Chris shares Eaton’s experience with MATLAB and Simulink version upgrades.
There are two important aspects in testing: test results and how much of the model and code is covered with executed test cases. Coverage is mandatory especially in safety-related projects. In this presentation, Mohammad introduces a new workflow that can be automated to prepare a Simulink® model as well as the generated code from the model using Embedded Coder®. This new workflow can be easily and directly integrated into an existing tool for code coverage capability.
The methodology gives engineers the ability to develop functional content using their choice of software tools (Simulink®, C code, etc.) and supports their feature-architecture level integration and vehicle-specific configuration in Simulink. These features are deployed on multiple platforms (e.g., dSPACE® RCP, PC simulation, production ECUs) using a custom code-generation framework.
This presentation shows you how engineers are developing advanced driver assistance systems using MATLAB® and Simulink®. Using an automatic emergency braking example, Abhishek talks about how you can:
- Reduce time on the road by simulating scenarios
- Speed up prototyping of algorithms by generating code
- Verify deployed algorithms by automating HIL testing
Learn a workflow to develop a real-time model that enables hardware-in-the-loop (HIL) testing of embedded motor controllers. This presentation demonstrates this workflow through a case study based on a permanent-magnet synchronous machine. The workflow takes into account:
- Modeling motor dynamics required for HIL testing
- Deploying the motor model to a HIL system
- Testing an embedded motor controller with the HIL system
There are different ways to perform optimization on a simulation model. Examples include parameter estimation and response optimization. In both cases, the objective function to be minimized is the result of the simulation. This presentation illustrates how you can:
- Leverage MATLAB® functionality from Simulink®
- Perform model correlation to obtain realistic parameter estimates for the plant model
- Perform design optimization on controller gains to follow dynamic response requirements
Verification and validation techniques applied throughout the development process enable you to find errors before they can derail your project. In this session, you’ll learn how to:
- Use Model-Based Design to perform virtual testing early in the project
- Verify and analyze your design to gain confidence that the design will meet specifications
- Reuse your tests on the code and verify your code to prove that is free of critical errors
ADAS and Autonomous Driving
Robot Operating System (ROS) is gaining popularity in autonomous driving development. With the available interface between MATLAB® and ROS, you can log data from sensors in the ROS environment and quickly postprocess the data in MATLAB for visualization and algorithm design. You can also deploy algorithms for path planning, computer vision, and control from MATLAB and Simulink® into the ROS environment. This presentation describes how the combination of MATLAB and ROS can help engineers in autonomous driving.
This session provides an example workflow for developing ADAS algorithms in MATLAB®. The example workflow is based on a vision and radar sensor fusion algorithm and demonstrates how you can:
- Gain insight by replaying and visualizing logged vehicle data
- Reduce time on the road by synthesizing sensor data to test algorithms
- Speed up prototyping of algorithms by generating code
The use of LiDAR as a sensor for perception in Level 3 and Level 4 automated driving functionality is gaining popularity. MATLAB® and Simulink® can acquire and process LiDAR data for algorithm development for autonomous driving functions such as free space detection. With point cloud processing functionality in MATLAB, you can develop algorithms for LiDAR processing, and visualize intermediate results to gain insight into system behavior. This talk shows how you can use 3D point cloud processing functionality to process LiDAR data for autonomous driving. Avi also shows how LiDAR processing can be combined with other common sensors such as cameras and radars.
Part of release R2016a, the MATLAB Live Editor provides a new way to create, edit, and run code. You can see results together with the code that produced them. Add equations, images, hyperlinks, and formatted text to enhance your narrative. Also, you can share with others as interactive documents. This presentation also provides an overview of other new capabilities beyond Live Editor.
Are you struggling with processing large amounts of data? Do you have difficulty making sense of the data you have? In this session, Will discusses data analytics and what it means to you. He presents different strategies to help you develop data analytics with MATLAB®.
This session shows how to use machine learning techniques in MATLAB® to identify anomalies and predict future engine health. Using data from a real-world example, Seth explores:
- Importing, preprocessing, and labeling data
- Selecting features
- Training and comparing multiple machine learning models
Learn how MATLAB is used to build prognostics algorithms and take them into production, enabling companies to improve the reliability of their equipment and build new predictive maintenance services.