Search Results

Originally developed for the service robot industry, the Robot Operating System (ROS) has lately received a lot of attention from the automotive sector with use cases, especially, in the area of advanced driver assistance systems and autonomous driving (ADAS/AD). Introduced as communication framework on top a of a host operating system, the value proposition of ROS is to simplify the software development in large-scale heterogeneous computing systems. Developers can focus on the application layer and let ROS handle the discovery of all participants in the system and establish communication in-between them. Despite the recent success of ROS, standardized automotive communication protocols such as the Universal Measurement and Calibration Protocol (XCP) are still dominant in the electronic control unit (ECU) development of traditional vehicle subsystems like engine, transmission, braking system, etc.

The powertrain module is being introduced to embedded System on Chips (SoCs) designed to increase available computational power. These high-performance SoCs have the potential to enhance the computational power along with providing on-board resources to support unexpected feature growth and on-demand customer requirements. This project will investigate the radial basis function (RBF) using the Gaussian process (GP) regression algorithm, the ETAS ASCMO tool, and the hardware accelerator Advanced Modeling Unit (AMU) being introduced by Infineon AURIX 2nd Generation. ETAS ASCMO is one of the solutions for data-driven modeling and model-based calibration. It enables users to accurately model, analyze, and optimize the behavior of complex systems with few measurements and advanced algorithms. Both steady state and transient system behaviors can be captured.

Data-driven plant models are well established in engine base calibration to cope with the ever increasing complexity of today’s electronic control units (ECUs). The engine, drive train, or entire vehicle is replaced with a behavioral model learned from a provided training data set. The model is used for offline simulations and virtual calibration of ECU control parameters, but its application is often limited beyond these use cases. Depending on the underlying regression algorithm, limiting factors include computationally expensive calculations and a high memory demand. However, development and testing of new control strategies would benefit from the ability to execute such high fidelity plant models directly in real-time environments. For instance, map-based ECU functions could be replaced or enhanced by more accurate behavioral models, with the implementation of virtual sensors or online monitoring functions.

The use of design of experiment (DoE) and data-driven simulation has become state-of-the-art in engine development and base calibration to cope with the drastically increased complexity of today's engine ECUs (electronic control units). Based on the representation of the engine behavior with a virtual plant model, offline optimizers can be used to find the optimal calibration settings for the engine controller, e.g. with respect to fuel consumption and exhaust gas emissions. This increases the efficiency of the calibration process and reduces the need for expensive test stand runs. The present paper describes the application of Gaussian process regression, a statistical modeling approach with practical benefits in terms of achievable model accuracy and usability. The implementation of the algorithm in a commercial tool framework enables a broad use in series engine calibration.