On the optimal parking problem (Q918494)

The authors study one of the well-known problems of stochastic optimal control, known as ``optimal parking''. This is a problem faced by a motorist in search of a parking space en route to his destination which is a random variable with a known distribution. The appearance of unoccupied parking spaces is also a random quantity, described by a Poisson processes. If the individual decides to take an available parking space, then it is assumed that he will walk the rest of the way to his destination, with the speed of walking being less than the speed at which the car can travel. The objective is to minimize the expected time it takes to reach the destination. The authors solve this problem by using the concept of dual predictable projection, and provide an explicit expression for the optimal stopping rule.