We introduce the physical puzzle Rush Hour and its generalization. We briefly survey its complexity limits, then we model and solve it using declarative paradigms. In particular, we provide a constraint programming encoding in MiniZinc and a model in Answer Set Programming and we report and compare experimental results. Although this is simply a game, the kind of reasoning involved is the same that autonomous vehicles should do for exiting a garage. This shows the potential of logic programming for problems concerning transport problems and self-driving cars.