Scalable Models Inspire Agile Development

Published: January 12, 2022

Ahmed Altarooti, Senior Application Engineer, Automated Driving & Software Solutions, dSPACE GmbH

Norbert Meyer, Manager Vehicle Simulation, Automated Driving & Software Solutions, dSPACE GmbH

Software development in the automotive industry increasingly relies on agile methods and continuous integration. Due to the special requirements of the code to be tested, imposed by factors such as real-time capability, demands on the quality and thus on testing are high.

This often requires new methods and processes to achieve functional test systems. The commissioning and deployment of test systems is undergoing a transformation from an expert discipline, where one step is performed at a time, to agile processes in which all steps, from the development of large simulation models to the deployment of multiple test systems, are performed in parallel by specialized teams.

New Model Structures Support Scalability and Agility in the Development Process

The structure of the model components of the Automotive Simulation Models (ASM) tool suite has been optimized for agile working, and the scalability of ASM simulations has been improved. The detailed selection of model parts was chosen in a way that enables easy integration and fast exchange. Generalized modules, such as vehicle dynamics, driver or engine modules, let you use decentralized model architectures so that simulations can be better parallelized. By standardizing interfaces, it also becomes easier to exchange information with different levels of detail, such as longitudinal dynamic driving resistance analyses with a multibody simulation of the complete vehicle dynamics, or the empirical electric motor simulation based on the torque requirement with the three-phase rotating field machine simulation. The newly created structures make it possible to scale ASM simulations in a targeted manner.

In addition, separating modules at logically decoupled interfaces lets you focus on individual functional parts on the computational task level. This results in a system setup that you can adapt to your computing power requirements, since model components are precompiled and can be arranged between computing tasks to improve performance.

Partitioning models into appropriately sized modules is key to successful system architectures.

The structure of the controller and controlled system in the modules in combination with the communication structure, allows entire modules to be interchangeable, which in turn makes it easy for you to split controller parts if the test system setup requires the outsourcing of controller parts.

Different workflows in in-the-loop projects were also analyzed in this regard before deciding on an ASM model structure. This made it possible to incorporate dSPACE's many years of experience into the advancement of the models.

The resulting structure thus combines the model development requirements for the test context with those of model distribution in the test setup and model connection to the control system.

More Information Product Information

Subscribe newsletter

Subscribe to our newsletters, or manage or delete your subscriptions