The importance of vehicle dynamics for ADAS and Autonomous Vehicle testing

There is a lot of effort being put into the development of virtual test environments for ADAS and AV. One aspect that seems to be ignored by many is the need for high fidelity road models and vehicle dynamics models. Whilst in the early stages of the development of ADAS and AV systems it may be acceptable to ignore these effects; they are significant and must be considered when validating the controllers. You need to be certain that your algorithms can cope with all the noise introduced by a real road surface.

To investigate this idea we constructed a detailed vehicle dynamics model using VeSyMA and Dymola. We then simulated this vehicle driving along 2 similar road models. The overall profile of the 2 roads is the same i.e. the centreline follows the same x, y, z coordinates but the surface roughness is very different. In the first example the road surface is perfectly smooth and in the second a more accurate model of the road surface is used that includes higher frequency effects and variations across the width of the road. The video below compares the car driving on both roads (the smooth road is on the left).

In the video the blue arrows are used to visualise the tyre forces. It’s clear that there is much more variation in the tyre forces on the detailed road and this will have an impact on the overall vehicle body position and orientation.

So, why is this important for ADAS and AV?

In these types of vehicles we have sensors that are rigidly mounted to the vehicle body and they need to look a long way ahead (and potentially all around) the vehicle. Let’s consider what happens to the points these sensors are looking at as we drive along the road. For example, if a sensor such as a radar is supposed to detect objects 50m ahead of the vehicle and up to 45 degrees away from the direction of travel, how different is the point it is looking at on these different roads?

The plot below shows the difference in position of this target point. At a range of just 50m we are looking at points that are 2.6m apart! We also see that the position we are looking at changes very quickly as the vehicle drives along the bumpy road. This could have a significant impact on our ability to accurately detect and classify objects.

In this example we have still ignored a lot of high frequency road surface imperfections that could have a further impact on where our sensors are looking. These effects can be added though as the road models in VeSyMA support adding roughness according to the ISO-8608 standard or we can use 2D tables of irregular roughness data and superimpose these onto a smooth road surface. We could also use high fidelity road models from rFpro which are built from LiDAR scans to millimetre precision and capture all the surface imperfections. As you increase the fidelity of the road surface though you also need to improve the fidelity of your tyre model to also support the increased frequency content of the road.

There are many reasons to make sure that your chosen virtual test environment supports high fidelity road and vehicle models, this is just one. If you would like to know more about how rFpro can be used for virtual testing or how Dymola and VeSyMA vehicle models can be used to add detail to your chosen environment then please get in touch.

Written by: Mike Dempsey – Managing Director

Please get in touch if you have any questions or have got a topic in mind that you would like us to write about. You can submit your questions / topics via: Tech Blog Questions / Topic Suggestion


Got a question? Just fill in this form and send it to us and we'll get back to you shortly.


© Copyright 2010-2024 Claytex Services Ltd All Rights Reserved

Log in with your credentials

Forgot your details?