Abstract. A* (a-star) is a well known best-first search algorithm that has been applied to the solution of different problems. In recent years, several extensions have been proposed to adapt it and improve its performance in different application scenarios. In this paper, we present a survey and classification of the main extensions to the A* algorithm that have been proposed in the literature. We organize them into five classes according to their objectives and characteristics: incremental, memoryconcerned, parallel, anytime, and real-time. For each class, we discuss its main characteristics and applications and present the most representative algorithms.

Simulation games present several challenges for computer controlled players. As a result of this, most of the artificial intelligence algorithms developed so far, especially for construction and management simulation games, do not give satisfactory results when compared to human performance. In this paper we develop an AI to control an agent of OpenTTD, a open-source clone of Transport Tycoon Deluxe, one of the premier construction and management simulation games. To do this, we create and adapt artificial intelligence techniques to allow their use in a dynamic, multi-agent strategic environment. Named trAIns, the developed AI constructs and manages railroad routes, the most challenging transport type in the game. Several qualitative and quantitative experiments comparing trAIns with another AI are performed, bringing very good results.