Model-Based Development of Seating Comfort System | US Based Tier-I Supplier

Customer:

Tier-1 supplier for automotive and industrial products.

Business Challenge:

• During the Model Based Development of their latest Seating ECU, our customer was confronted with a unique challenge related to controller memory.
• For this project, the customer created floating point library model which occupies a large part of on-chip memory, because of the ‘double’ data type.
• Since this new Automotive Seating ECU featured some very advanced seating comfort and control functions, floating point library modeling turned out to be an inefficient method with respect to on-chip memory consumption
• To resolve this issue within time and budget constraints, the customer was on a look out for an embedded software partner with in-depth expertise in Model Based Development.

Embitel Solution:

• During the technology workshops, our automotive software development team and the customer agreed to convert the floating point models into fixed point models.
• Since there were multiple blocks of models that required the conversion, our embedded software developers also started working on the automation tool to make the entire process more efficient.

Step-by-step details of migration to fixed point models:

• The first step was to determine the data types of the signals and parameters for all the library models.
• Then the model initialization files were created, containing all the signals and parameters with all their required attributes.
• The fixed point models were created from provided floating point library models. Then, the optimized codes were generated.
• We also created interface files for low-level driver.
• Then, these MATLAB generated code files were integrated with the low-level drivers
• Our automotive software team leveraged the in-house designed automation tool for all the future releases
• Due to Automation of the whole process, the development time was reduced from several days to a 2-3 day process.
• The tool ensured automation of the following steps:
  • Model Initialization file creation
  • Interface Files Creation
  • Fixed point code based Model creation
  • Code generation

Embitel Impact:

• Our automotive software development team’s pro-active approach helped us to identify the scope of automation. This ensured reduction in the turnaround time for the new releases.
• The automation tool reduced the overall process time by around 80% and has substantially improved the accuracy of the work products.
• As per the client requirement, with minimal changes, the tool can be reused across multiple MATLAB/Simulink projects.

Tools used:

• Matlab / Simulink 2013b
• Matlab / Simulink 2016b
• ERT coder
• IAR compiler
• CANoe