Unmanned Cluster Intelligence Level Evaluation Based on ELO Scores and Generative Adversarial Network
Bo Fan, Xiaoxuan Lv, Jilong Zhong, Lixia Xu, Shaoshi Wu, Yishan Ding, Xiaoyu Zhai, Xinwen Hou · 2023
The evaluation of intelligence levels is crucial in assessing the capabilities of unmanned clusters. Currently, this process is often determined subjectively, leading to potential biases and missing evaluation indicators due to limited simulation experiments. To address these issues, this paper introduces an objective method for determining intelligence levels using the ELO rating algorithm and constructs an intelligence level evaluation model based on the Generative Adversarial Network (GAN) to account for missing evaluation indicators. The model takes evaluation indicators as input and ELO scores as output, providing an objective reflection of the intelligence capabilities of unmanned clusters. Using an airground collaborative encirclement task as an example, the model is trained to evaluate the intelligence levels of unmanned clusters objectively. Simulation results demonstrate the method's effectiveness in quantifying intelligence levels objectively and its applicability to a wide range of unmanned clusters.