This paper introduces a novel pattern recognition method called k-SVM, which combines k-means clustering with Support Vector Machines (SVM) to enhance the classification of robot types in collaborative tasks. The method first uses k-means clustering to group robot data into distinct clusters, reducing the number of support vectors and improving feature discrimination. Then, SVM is applied to construct a discriminative hyperplane for classification. The k-SVM approach is validated through cross-validation experiments, demonstrating superior performance in robot classification compared to traditional SVM methods. The method is particularly effective in handling non-linearly separable data and improves classification accuracy and speed. The experiments show that k-SVM outperforms traditional SVM in recognizing diverse robot patterns, achieving lower classification errors as the dataset size increases. The results indicate that k-SVM is more efficient and accurate in classifying robot types, especially in dynamic and complex environments. The method is applied to distinguish between flying and mobile robots in collaborative tasks, with clear separation between the two classes. The proposed k-SVM method is shown to be effective in improving the classification of robot patterns, enhancing the efficiency and accuracy of multi-robot systems.This paper introduces a novel pattern recognition method called k-SVM, which combines k-means clustering with Support Vector Machines (SVM) to enhance the classification of robot types in collaborative tasks. The method first uses k-means clustering to group robot data into distinct clusters, reducing the number of support vectors and improving feature discrimination. Then, SVM is applied to construct a discriminative hyperplane for classification. The k-SVM approach is validated through cross-validation experiments, demonstrating superior performance in robot classification compared to traditional SVM methods. The method is particularly effective in handling non-linearly separable data and improves classification accuracy and speed. The experiments show that k-SVM outperforms traditional SVM in recognizing diverse robot patterns, achieving lower classification errors as the dataset size increases. The results indicate that k-SVM is more efficient and accurate in classifying robot types, especially in dynamic and complex environments. The method is applied to distinguish between flying and mobile robots in collaborative tasks, with clear separation between the two classes. The proposed k-SVM method is shown to be effective in improving the classification of robot patterns, enhancing the efficiency and accuracy of multi-robot systems.