Putting ice on the Road.

[OlivertheMattes]

Good Morning Everyone,
I am currently preparing for a Research Project about Reinforcement Learning.
Specifically, I want to do work about Concept Drift detection and handling.
To that End, I need a Machine Learning Problem with Concept Drift.
My Idea was, to use Torcs as an Environment. The Machine Learning Algorithm would then Learn the Central concept of gameplay, how to controll the Car.
This, i already know is possible.
However, there is no Concept Drift in this Problem, because the behavior of the Car does not change, while on the road.
My idea was to introduce Concept drift by having the Road slowly ice over . Not visually, of course, only in regard to the way the Car behaves on the track.
Is there any way to do this in TORCS?
Thank you for your time.