Analysis of Robocode Robot Adaptive Confrontation Based on Zero-Sum Game
This addresses adaptive strategy development in simulated robot confrontations, but it is incremental as it applies known algorithms to a specific gaming domain.
The paper tackled the problem of adaptive confrontation in non-complete information environments by testing a Zero-sum Game Alpha-Beta pruning algorithm combined with opponent motion prediction and early bullet release in Robocode tank robots. The result involved experimental confrontations among seven robots, with effectiveness verified through histograms and radar plots to express tank intelligence differences.
The confrontation of modern intelligence is to some extent a non-complete information confrontation, where neither side has access to sufficient information to detect the deployment status of the adversary, and then it is necessary for the intelligence to complete information retrieval adaptively and develop confrontation strategies in the confrontation environment. In this paper, seven tank robots, including TestRobot, are organized for 1V 1 independent and mixed confrontations. The main objective of this paper is to verify the effectiveness of TestRobot's Zero-sum Game Alpha-Beta pruning algorithm combined with the estimation of the opponent's next moment motion position under the game round strategy and the effect of releasing the intelligent body's own bullets in advance to hit the opponent. Finally, based on the results of the confrontation experiments, the natural property differences of the tank intelligence are expressed by plotting histograms of 1V1 independent confrontations and radar plots of mixed confrontations.