Skip to main content

Lars Svensson Tests Traction Adaptive Motion Planning in Self-Driving Trucks

See the new video

Published Jun 17, 2020

ITRL researcher Lars Svensson has been conducting some exciting evaluations of his research on motion planning and control for self-driving vehicles. See the video and learn more about his project below!

What is the project about?

We have developed a framework for motion planning and control for selfdriving vehicles that takes road conditions, e.g. traction, into account. By doing that the full physical capacity of a vehicle can be used to avoid accidents in critical situations, both when road conditions are good and poor.
 
Why is this good?

Well, when traction is really good, the framework will plan and execute very aggressive motions to avoid collision, for example turning and braking hard. On the other hand, when traction is poor (as in the video), the framework will plan and execute smoother motions, that still utilizes all available grip, but avoids loss of control due to excessive tire slip. In both cases, traction adaptation increases the probability to avoid or mitigate accidents in critical situations.

And now you have tested the algorithms in real trucks?

Yes, we have been collaborating with the Revere Lab at Chalmers  to evaluate the algorithm in realistic conditions using a full scale heavy duty vehicle. We have tested it on various road surfaces like wet/dry asphalt at the Asta Zero  test facility, and on a special low friction surface at the Stora Holm driver's education facility .

So how does it work? Is it a secret?

No definitely not. The technical details of the core concept are published in a paper available here

We are currently working on an extended version including the experimental results, that will be available soon.

How can I find out more?

For further questions, contact Lars