Yes, at least some drivers deliberately try to kill animals. I’ve seen a sports car change lanes intentionally to run over and kill a Golden Retriever puppy (angriest I’ve ever been at a driver).
I think it may be eye-opening when self-driving vehicles with collision avoidance become the norm. We might see roadkill almost disappear, implying that human behavior (either murderous or inattentive) has been the problem all along, not animal behavior.
Naw, I hit a raccoon once. There was absolutely no way I could have avoided it, it ran in front of me on a road too narrow to safely swerve.
There’s a reason that insurance policies treat “I struck a wild animal” as an act of God, or as something you have no control over. A large fraction of animal strikes are the fault of the animal, or just bad luck.
Self driving cars might attempt to avoid hitting animals, but not if that produces more risk to the human occupants. For animals raccoon size or smaller it might be safer to hit them than to swerve.
What if the best instinct for avoiding being roadkill is at odds with the best instinct for predator evasion? I’ve seen more than a few squirrels race across a road, just barely clear the second tire, then suddenly double back and get hit. I assumed that the shadow of the car caused them to jink backwards, which would work for predators attacking from the air, but not for dealing with cars.
In order for a characteristic to be selected for it has to be present in the population, and the prevalence of the characteristic at the time the selective pressure is applied will have a strong effect on how quickly the characteristic achieves dominance. For example in the case of Squirrels turning black, due to just general diversity among squirrel coats, black squirrels may have represented 10% of the population before the coal era. If being black improved your likelihood of successfully reproducing by 5%, than in only 60 generations you would have approximately 2/3 of the squirrels being black. But I would expect that a complicated characteristic of “look both ways before crossing the street”, or avoid big black strips, would be much less likely to exist in the random population. If such was present in 1/10,000 squirrels, and had the same 5% advantage to reproduction, it would represent only one about 450 squirrels by the 60th generation. It would take about 140 generations, it will be up to 10% of the population, and then be up to 2/3 another 60 generations after that.