User Stories

LS Automotive Reduces Development Time for Automotive Component Software with Model-Based Design

Challenge

Shorten development times for embedded control software used in automotive switches and components

Solution

Use Model-Based Design to model controller designs, run simulations, verify customer specifications, and generate error-free production code

Results

  • Specification errors detected early
  • Proven development approach established
  • Coding errors eliminated

“By enabling us to analyze requirements quickly, reuse designs from previous products, and eliminate manual coding errors, Model-Based Design has reduced development times and enabled us to shorten schedules to meet the needs of our customers.”

MyoungSuk Ko, LS Automotive

An LS Automotive door area unit.


One of the largest tier 1 suppliers of vehicle switches and components in Korea, LS Automotive continues to win contracts from both domestic and overseas automotive OEMs. There is an increasing demand from those OEMs for faster delivery of more features and for production systems that comply with the ISO 26262 standard.

To meet these demands, LS Automotive worked with MathWorks Consulting Services to adopt Model-Based Design with MATLAB® and Simulink®. On their first project with Model-Based Design, LS Automotive engineers modeled, simulated, and generated embedded software for a door area unit (DAU), which controls the vehicle’s power windows and mirrors.

“Our team is tasked with developing many projects—in addition to the DAU—with limited human resources,” says MyoungSuk Ko, senior engineer for the software design team at LS Automotive. “Model-Based Design enables us to overcome this limitation through model reuse, early verification of customer requirements, and the elimination of human errors that are common in manual coding. Model-Based Design is especially beneficial for back-to-back testing in the context of ISO 26262—it enables our customer to ensure consistency between the requirements models and the generated C code.”

Challenge

Formerly, LS Automotive engineers used a traditional development process in which embedded software was written by hand. Even with a common set of specifications, the handwritten code often varied from developer to developer. In addition, human errors—in the code or in the customer’s specification—caused chronic quality problems that had to be remedied in the later stages of development.

In addition to these challenges, LS Automotive recognized the need to respond to the increased demand from overseas OEMs for products developed with model-based approaches rather than with a traditional approach. The application of ISO 26262 both in Korea and globally underscored the need for compliance with functional safety standards. As part of their functional safety requirements, global automotive OEMs often requested a simulation-based verification environment that was independent of the actual hardware.

LS Automotive needed to meet these demands while addressing the limitations it had identified in its existing development process.

Solution

LS Automotive adopted Model-Based Design and reduced risk by engaging MathWorks Consulting Services to review initial models and advise on functional behavior testing.

Working in Simulink and Stateflow®, LS Automotive engineers modeled the DAU’s mirror control, power window control, integrated memory system, and switch input/output modules based on specifications provided by the OEM.

They ran simulations in Simulink to test each module independently. During the simulations, and with support from MathWorks consultants, the team measured model coverage with Simulink Coverage™ and applied formal methods to detect dead logic in their models using Simulink Design Verifier™.

The simulations revealed customer specification errors, but these were relatively easy to resolve because they had been identified early in development.

With support from MathWorks consultants, the team used Simulink Test™ to create test harnesses, implement test scenarios, manage and execute test suites, and analyze test results. Following test result analysis, they used Simulink Design Verifier to generate test case inputs that maximized model coverage.

After validating the specification and completing the functional verification of the design in Simulink, the team generated about 6000 lines of C code from their models using Embedded Coder®. They compiled the code for the target MPC5601D microcontroller using Freescale™ CodeWarrior® embedded software development tools.

Development of the DAU is on schedule, and the company expects a volume of 2 million units in Korea and North America when the part moves to mass production. LS Automotive engineers are now applying Model-Based Design to a power seat module project, and they plan to expand its use to upcoming steering angle sensor and body control module projects.

Results

  • Specification errors detected early. “Model-Based Design has enabled us to ensure quality while reducing development schedules,” says Ko. “We analyzed requirements to quickly find potential errors. We verified algorithms early in development—before the actual hardware was available—and reused 80% or more of our models.”
  • Proven development approach established. “Within our organization, many people initially voiced concerns about using an approach and tools that we lacked experience with on a mass production project,” Ko says. “Now, thanks to the efforts of MathWorks consultants on the DAU, there is a strong voice calling for Model-Based Design to be used on more products.”
  • Coding errors eliminated. “By using Embedded Coder to generate code, we have eliminated the human errors that were common in hand code,” says Ko. “This, combined with other quality improvements enabled by Model-Based Design, helped DAU become our first product to pass initial QA tests with no flags.”