For a better experience on, enable JavaScript in your browser. Thank you!

Virtual Test Drives Are Key to Testing Functions for Highly Autonomous Driving

Published: February 28, 2017

dSPACE Tech Day on February 23, 2017, Novi, MI

Automotive OEMs are pushing forward and taking the lead in this exciting new era of connected and automated driving, but a lot of research still needs to be conducted. In fact, it will take millions of scenarios, matched against tens of thousands of requirements and tests, before the first fully automated vehicle can be safely launched into the marketplace.

But how can you effectively carry out and manage the enormous number of required tests?

More than 235 professionals from the engineering community learned that the answer lays with simulation-based development at a dSPACE Tech Day event held on Feb. 23, 2017, in Novi, Michigan, USA.

Virtual Test Drives – A Better Approach

dSPACE technical experts led a series of presentations and focus workshops to educate attendees on steps they can take to develop and validate functions for highly autonomous driving using dSPACE’s comprehensive tool chain.

“Experts say that highly automated vehicles will have to be test-driven hundreds of millions of miles, but obviously, it is unpractical to do this by on-road driving alone,” said André Rolfsmeier. “A better approach is to implement virtual test drives.”

Using a combination of model-in-the-loop (MIL) and software-in-the-loop (SIL) simulation of control algorithms with realistic virtual models of vehicle traffic and the surrounding environment, you can cover a broad range of test drives that factors in all kinds of scenarios … lane detection, object detection, traffic sign recognition, lane keeping assist, parking assist, road types, weather, traffic scenarios, etc.

Rolfsmeier presented a video (Highly Automated Driving) of a simulated vehicle driving through an urban area including more than  120 dynamic objects and sensors such as a front radar sensor for object detection and a front camera sensor for lane detection.

Instead of waiting for the real ECU prototypes to become available, Rolfsmeier said you can save so much time and make a huge jump start on testing and validating ECUs using the PC-based simulation platform VEOS during the early development phase.

Test scenarios with VEOS are completely automated. You can set up as many VEOS test cases in one cluster as needed for parallel execution of virtual test drives.  Using virtual ECUs, together with VEOS, makes for a highly scalable test system. 

dSPACE Tool Chain Related to Autonomous Driving

While MIL/SIL represents an important step in the testing and validation process, it is only one part of the overall development and testing activity. Rolfsmeier shared dSPACE’s complete tool chain for autonomous driving development.

dSPACE’s strategy for achieving highly autonomous driving is a complete and powerful platform that can handle every development stage:

  • Prototyping tasks, including the development of control algorithms for real-time sensor applications
  • MIL/SIL simulation to create virtual ECUs that can enable fast and reproducible test execution at an early development stage for virtual test drives
  • Production code generated directly from Simulink® / Stateflow® models
  • HIL simulation to perform automated software tests and simulation drives for sensors, data fusion, image processing ECUs, radar- and camera-based ADAS, V2X applications based on DSRC, etc.
  • Real test drives with HIL system sensor integration and data logging options

The Need for a Unified Architecture

At the Tech Day event, Prof. Dr. Levent Guvenc from the Ohio State University reminded everyone that there are many players in the connected and automated driving arena and most are using proprietary solution architectures. He emphasized the importance of establishing a unified architecture that is both scalable and replicable.

Dr. Guvenc presented case studies for the development of autonomous features and extensive use of dSPACE tools.

At the OSU Automated Driving Lab, Guvenc and his core research group (comprised of faculty members, researchers and graduate students) are developing connected and automated driving vehicles, including low-speed autonomous shuttles for smart cities. For these projects, OSU is utilizing a dSPACE prototyping unit (MicroAutoBox) and a dSPACE simulator to evaluate and test controls for automated driving functions. Additionally, they are using dSPACE TargetLink for production code generation.

Support for Electric Motor Applications

Electrification was another key topic of discussion at the dSPACE Tech Day. Three workshops were held on this topic, covering solutions from smart grids and energy storage to motor controls development and testing.

More Information