Human motor planning involves, at least, forward and inverse models for performing stochastic optimal control (at least, that's the standard formulation of the problem). An intuitive physics engine may be built into our heads, but whether or not it is, just running it forward doesn't seem to suffice for motor control.