Abstract
A basic routing algorithm is presented for finding optimal routes for cars in a rapid transit system. The algorithm uses predicted path cost, depth-first search, and threshold acceptance to minimize computation cost. It is applicable to synchronous, cycle-synchronous, and trans-synchronous control strategies, and will minimize either departure time, transit time, or arrival time as may be needed. Extensions of the algorithm to allow finding empty cars to answer service requests, to dynamically reroute cars in the system, and to handle multistation routes are presented.