The study proposes a rotational multipyramid network (RoMP Net) with bounding‐box transformation for object detection. The RoMP Net is a single‐stage object detection neural network featuring three characteristics. First, the network uses a rotational bounding box to minimize the effect of background images when extracting features of objects. Bounding‐box transformation was proposed to compensate for the limitation of the rotational bounding boxes, which have relatively low prediction accuracy for objects with a high aspect ratio. Second, the RoMP Net introduces a multi‐scale and multi‐level feature pyramid network to extract distinct and semantic features efficiently. This network architecture ensures high prediction accuracy and robustness regardless of the size and complexity of objects. Third, hyperparameters in the bounding boxes are automatically determined through an unsupervised clustering method. This optimization method is also critical in improving accuracy. The performance of the proposed network and preprocessing methods are validated through image‐sets comprising critical components in power transmission facilities, which have a variety of sizes and aspect ratios. This case study demonstrates the effectiveness and robustness of the three key characteristics in the RoMP Net. Furthermore, the RoMP Net outperforms other state‐of‐the‐art deep neural networks in prediction accuracy and robustness for object detection. Specifically, the mean average precision of the RoMP Net in the validation image‐sets shows that it has the highest prediction accuracy, whereas its values in the test image‐sets confirm the network's robustness. The fast yet accurate RoMP Net will expand the range of object detection through deep neural networks.

No abstract

Vibration isolation with mode decoupling plays a crucial role in the design of an intelligent robotic system. Specifically, a coupled multi-degree-of-freedom (multi-DOF) model accurately predicts responses of system dynamics; hence, it is useful for vibration isolation and control with mode decoupling. This study presents a vibration isolation method with mode decoupling based on system identification, including a coupled multi-DOF model to design intelligent robotic systems. Moreover, the entire procedure is described, including the derivation of the governing equation of the coupled multi-DOF model, estimation of the frequency response function, and parameter estimation using least squares approximation. Furthermore, the suggested methods were applied for a mobile surveillance system suffering from resonances with mode coupling; it made the monitoring performance of the surveillance camera deteriorate. The resonance problem was mitigated by installing vibration isolators, but limited to eliminate the coupling effects of natural frequency deterioration performances of vibration isolation. More seriously, system identification with a simple decoupled model limits the prediction of this phenomenon. Hence, it is difficult to enhance the performance of vibration isolators. In contrast, the presented method can accurately predict the vibration phenomenon and plays a critical role in vibration isolation. Therefore, dynamic characteristics were predicted based on a vibration isolator using the coupled three-DOF model, and a final suggestion is presented here. The experiments demonstrated that the suggested configuration decreased vibration up to 98.3%, 94.0%, and 94.5% in the operational frequency range, i.e., 30–85 Hz, compared to the original surveillance system in the fore-after, side-by-side, and vertical directions, respectively. The analysis suggests that the present method and procedure effectively optimize the vibration isolation performances of a drone containing a surveillance system.