Anomaly detection is the cornerstone of the health management of many large industrial mechanical equipment. Most machinery anomaly detection methods try to find a variable threshold to indicate whether machinery is functioning normally. However, due to the scarcity of anomalous data, how to drive anomaly detection with variable threshold only by normal data remains a significant challenge. This is because, in practice, mechanical equipment’s operating speed and load conditions are constantly changing. To solve this problem, an unsupervised anomaly detection approach based on autoencoders and temporal convolutional networks was proposed in this paper, which is named Variable Cumulative Error Anomaly Detection (VCEAD). The autoencoder will be used to compute the signal reconstruction error, and temporal convolutional network will be used to predict vibration signal. Finally, an algorithm based on the probability distribution of the signal reconstruction error will be proposed to calculate the so-called variable cumulative error. To validate the effectiveness of the proposed method in anomaly detection, multi-group lifetime datasets collected from the laboratory were studied and analyzed. The comparative results show that the method presented in this paper achieves superior performance.

Tendon-driven continuum robots (TDCRs) are of great promise in dexterous manipulation in long-narrow spaces, such as in-situ maintenance of aero-engines, due to their slender body and compliant hyper-redundant architecture. However, major challenges in implementing this come from mechanical design and morphology estimation: (1) torsion and buckling issues induced by the intrinsic compliant architecture and the coupling of system gravity and distal loads; (2) low-accuracy morphology model influenced by complex load conditions. In this article, inspired by the contact-aided compliant mechanisms (CACMs), a novel continuum robotic system using the bearing-based CACM is developed to overcome the two intrinsic issues (i.e., torsion and buckling) while eliminating the implied wear due to friction at joint/socket interfaces without affecting its stiffness adversely. Subsequently, based on the chained beam constraint model, a comprehensive kinetostatic modeling framework is systematically derived, focusing on mechanism-oriented strategies (i.e., tendon routing friction, physical joint constraint, and section buckling estimation). Finally, various experiments are performed to verify the effectiveness of both our designed hardware and algorithm. It is demonstrated that the robotic system with such hardware and algorithm achieving the torsional stiffness outperforms the twinpivot design at least 24 times, stiffness enhancement > 100 times, morphology error < 2.5% of the manipulator length, and avoiding the first-order instability. Additionally, we demonstrate the navigation experiment by using two developed control strategies to show the performances of the robotic system.

Drawing inspiration from the jumping mechanisms of insects (e.g., click beetles), bistable structures can convert slow deformations of soft actuating material into fast jumping motions (i.e., power amplification). However, bistable jumpers often encounter large energy barriers for energy release/re‐storage, posing a challenge in achieving multimodal (i.e., height/distance) and continuous jumps at the insect scale (body length under 20 mm). Here, a new offset‐buckling bistable design is introduced that features antisymmetric equilibrium states and tunable energy barriers. Leveraging this design, a Boundary Actuation Tunable Energy‐barrier (BATE) jumper (body length down to 15 mm) is developed, and transform BATE jumper from height‐jump mode (up to 12.7 body lengths) to distance‐jump mode (up to 20 body lengths). BATE jumpers can perform agile continuous jumping (within 300 ms for energy release/re‐storage times) and real‐time status detection is further demonstrated. This insect‐level performance of the proposed BATE jumper showcases its potential toward future applications in exploration, search, and rescue.

Capacitive sensors show promising applications for human–computer interface due to their potential capabilities of both pressure and proximity sensing. However, the sensitivity loss at large strain and the complicated process of fabrication technology limit related applications. Herein, inspired by gradient and hierarchical structures in biological systems and using porous materials with different stiffness, a capacitive flexible sensor with both proximity and pressure sensing abilities is proposed. Such hierarchical structures effectively reduce the attenuation of sensitivity as the pressure increases, realizing an ultrawide detection range with high sensitivity. A theoretical mathematical model is developed and the corresponding capacitive sensor is fabricated by using the high‐performance porous impregnated dust‐free paper as the electrode and expandable polyethylene (EPE)/loofah‐like polyurethane sponge (LPS) as double‐layer porous dielectric. This sensor has an ultrawide detection range up to 3000 kPa with a significantly low sensitivity loss of 0.0195% kPa⁻¹ and the noncontact sensing mode reaches a sensing range up to 28 cm (0.122 cm⁻¹). Notably, a strategy is provided to design the high sensitivity sensor over an ultrawide detection range from approaching to pressuring, and eventually its promising application is presented as human–computer interaction.

Condition monitoring (CM) of rotating machinery is crucial for condition-based maintenance, ensuring safety and timely fault detection. Principal Component Analysis (PCA), an unsupervised learning algorithm for data dimensionality reduction, is widely employed for feature extraction in CM. In contrast to typical applications where PCA serves as a data preprocessing tool, this paper proposes a novel approach that links principal components obtained through PCA dimensionality reduction of degraded data spectra with a data fusion level Health Indicator. The approach elucidates the physical significance and changing patterns of these principal components in the frequency domain. Notably, the second Principal Component-based Fourier Spectrum (PCBFS-2) effectively discriminates between normal and fault frequencies, enabling automated fault feature recognition. The effectiveness of this method in CM and fault feature recognition is validated through simulations and experiments. Importantly, this approach is interpretable, offering a new perspective for combining data-driven modeling with machine learning techniques.