无人集群已经广泛渗透并深刻影响到了现代社会的各个领域。在任务执行过程中，无人集群可能面临多种干扰和破坏，这会影响集群的正常运行及任务的高效完成。开展集群鲁棒性评估的研究，对于提升集群在动态环境下的行为协同稳定性与任务完成高效性，具有重要的理论意义与应用价值。本文以无人机集群协同感知为应用场景，根据场景任务中的协同行为特性，采用鹰鸽演化博弈建模集群协同过程，从集群属性、环境属性和任务效果三个维度构建无人机集群鲁棒性指标集，提出了鲁棒性量化指标，分析了集群在不同攻击场景下的鲁棒性。最后，将沙普利加和解释（SHapley Additive exPlain, SHAP）与极端梯度提升树（eXtreme Gradient Boosting, XGBoost）模型相结合，构建具有可解释性的鲁棒性智能量化评估模型。实验结果表明，本文提出的鲁棒性评估方法具有良好的可行性和有效性，提升了评估模型的透明度，能够为无人集群优化设计提供有效反馈。

Unmanned swarms have widely penetrated and profoundly affected all fields of modern society. In the process of mis-sion execution, unmanned swarms may face many kinds of interference and damage, which will affect the normal operation of the swarm and the efficient completion of the mission. Evaluating of the robustness of swarm has im-portant theoretical and applied value to ensure the stability and efficiency of the swarm in the dynamic environment. In this paper, taking cooperative sensing of the UAV swarm as an example, we utilize the Hawk and Dove evolutionary game to model the cooperation process according to behavioral characteristics of swarm cooperation. We construct the robustness indicator set of UAV swarm from the three dimensions of swarm attributions, environmental attributions and task effects and present a quantitative metric of evaluating robustness. Then we analyze the robustness perfor-mance of the swarm under various perturbation factors. Finally, we propose an intelligent evaluation model eXGBoost based on the Shapley additive explanation (SHAP) and the eXtreme Gradient Boosting (XGBoost) model. The exper-imental results show that the robustness indicator set and the evaluation method proposed by this paper have good feasibility and effectiveness. In addition, the method increases the transparency of the evaluation model. It can pro-vide effective feedback for the design of unmanned swarms. the design of unmanned swarms.