Abstract
The success of a Real-Time Strategy agent is heavily dependent on its ability to respond well to a large number of diverse tactical situations. We present a novel method of tactical decision making called Search and Recall (S&R) which is a hybrid of Search and Case Based Reasoning (CBR) methods. S&R allows an agent to learn and retain strategies discovered over the agent's history of play, and to adapt quickly in novel circumstances. The sense of memory that S&R provides an RTS AI agent allows it to improve its performance over time as better responses are discovered. S&R demonstrates an minimum win rate of 92% in standard scenarios evaluated in this paper. S&R decouples search from the main game loop which allows arbitrary computational complexity and execution time for search simulations. Meanwhile in-game decision making is based on CBR and remains fast and simple. This paper presents an S&R model which extends the ability of an RTS AI agent to deal with complex tactical situations. These situations include special unit abilities, fog of war, path finding, collision detection and terrain analysis.
Original language | English |
---|---|
Title of host publication | AISB Convention 2015 |
Place of Publication | United Kingdom |
Publisher | The Society for the Study of Artificial Intelligence and the Simulation of Behaviour |
Pages | 1-6 |
Number of pages | 6 |
Publication status | Published - 2015 |
Event | Artifical Intelligence and Simulation of Behaviour Convention - University of Kent, Canterbury; United Kingdom, United Kingdom Duration: 20 Apr 2015 → 22 Apr 2015 |
Conference
Conference | Artifical Intelligence and Simulation of Behaviour Convention |
---|---|
Country/Territory | United Kingdom |
Period | 20/04/15 → 22/04/15 |