The dog is doing this intuitively. There is no "dV/dT" calculation going on, the dog is simply estimating (based on experience) where the ball (or Frisby, or stick) is going to be at a particular time, and speeds up, slows, or turns to meet the object at that particular point. As the dog has its eye on the object, it continues to adjust the estimate to meet the anticipated contact point at the optimal time.
Human beings do the same thing when trying to beat out a red light.
How do you apply the statistical variable on the probability of a cop being near enough to observe the results and act accordingly? :-)
One shouldn't ascribe to mathematics that which instinct and common sense can achieve in 1/10000 of the time.