In this paper, a feature extraction approach for the deformable linear object is presented, which uses a Bezier curve to represent the original geometric shape. The proposed extraction strategy is combined with a parameterization technique, the goal is to compute the regression features from the visual-feedback RGB image, and finally obtain the efficient shape feature in the low-dimensional latent space. Existing works of literature often fail to capture the complex characteristics in a unified framework. They also struggle in scenarios where only local shape descriptors are used to guide the robot to complete the manipulation. To address these challenges, we propose a feature extraction technique using a parameterization approach to generate the regression features, which leverages the power of the Bezier curve and linear regression. The proposed extraction method effectively captures topological features and node characteristics, making it well-suited for the deformation object manipulation task. Large mount of simulations are conducted to evaluate the presented method. Our results demonstrate that the proposed method outperforms existing methods in terms of prediction accuracy, robustness, and computational efficiency. Furthermore, our approach enables the extraction of meaningful insights from the predicted links, thereby contributing to a better understanding of the shape of the deformable linear objects. Overall, this work represents a significant step forward in the use of Bezier curve for shape representation.

翻译：本文提出了一种面向可变形线性物体的特征提取方法，该方法采用贝塞尔曲线表征原始几何形状。所提出的提取策略与参数化技术相结合，旨在从视觉反馈的RGB图像中计算回归特征，最终在低维潜在空间中获取高效的形状特征。现有文献往往难以在统一框架中捕捉复杂特性，且在仅依赖局部形状描述子引导机器人完成操作任务时表现不佳。为应对这些挑战，我们提出了一种利用参数化方法生成回归特征的特征提取技术，该技术充分发挥了贝塞尔曲线与线性回归的优势。所提出的提取方法能有效捕获拓扑特征与节点特性，使其特别适用于变形物体的操作任务。通过大量仿真实验对所提方法进行评估，结果表明本方法在预测精度、鲁棒性及计算效率方面均优于现有方法。此外，本方法还能从预测链路中提取有意义的见解，从而增进对可变形线性物体形状的理解。总体而言，本研究为利用贝塞尔曲线进行形状表征开辟了重要进展。