Multimodal grasping has emerged as a promising strategy to enhance the grasping diversity of grippers in response to the rapid expansion of application scenarios. Among various designs, the pinch-suction hybrid mechanism and the soft-rigid hybrid structure have proved to be two practical strategies to achieve multimodality. However, existing research on these two strategies still lacks simple and effective collaborative mechanisms to fully leverage the advantages of each mode while ensuring mutual non-interference. In this article, we propose a pinch-suction and soft-rigid hybrid multimodal gripper (HMG), integrating four operating modes into a compact structure. Two simple and effective collaborative mechanisms are introduced to coordinate between pinch and suction operation and between soft and rigid components, respectively. Through the collaboration of different modes, the HMG exhibits a competitive grasping diversity across four aspects, including weight (from 0.2 g to 10 kg), fragility (from jelly to aluminum profile), size scale (from 0.46 mm to 0.55 m), and shape (from poorly pinchable to poorly suckable). We further demonstrate its adaptability and robustness in handling irregular-shaped objects, and its proficiency in executing complex real-world manipulation tasks, underwater operations, and closed-loop grasping. Its enhanced grasping diversity is poised to accelerate diverse applications in daily life, industrial settings, and underwater scenarios.

This paper tries to solve the challenges in few-shot transfer learning for anomaly detection: how to guarantee the transfer effect on insufficient even limited source domain data, and how to make the transfer process interpretable for getting trustworthy results. This paper proposes a deep domain-adversarial contrastive model with time-frequency transferability analytics. The essential idea is extracting fine-grained information from different frequency bands for reliable transfer. First, a time-frequency domain feature pool is constructed by applying wavelet scattering network (WSN) under different decomposition scales and rotation orientations. An orientation-first selection strategy is further designed to determine the optimal features that can cover the low, medium and high frequency bands. A new transferability metric, named frequency importance metric (FIM), is then built through frequency hypersphere matching to quantify the significance of each frequency band from source domain data. Second, a deep domain-adversarial contrastive network (DDCN) is constructed to realize selective information transfer according to frequency band’s significance. In DDCN, a purposeful feature representation can be extracted through the contrastive learning between the deep features and wavelet features that are weighted by FIM, thus leading to valid transfer in few-shot scenario via domain-adversarial training. Experiments are conducted on two typical anomaly detection problems, i.e., image recognition detection on the MNIST USPS and Office-Home datasets, and early fault detection on the IEEE PHM Challenge 2012 bearing dataset. The results not only verify the superior performance of the proposed method to the few-shot transfer learning, but also reveal the frequency saliency in the transfer process.

Fuzzy Petri nets (FPNs) are a vital modeling technique for the construction of knowledge-based systems, which have been commonly used in many fields, such as fault diagnosis, risk assessment, workflow management, and disassembly process planning. However, the conventional FPNs have been blamed for the following reasons: 1) the representation parameters in FPNs cannot precisely model experts' experience since it is difficult to manage the fuzziness and randomness of knowledge assessments simultaneously, and 2) the weight coefficients in the existing approximate reasoning algorithms are hardly enough to reflect the associated weights of reordered places. In response, we propose a new type of FPNs, called cloud reasoning Petri nets (CRPNs) based on the concept of interval clouds and the hybrid averaging operator. The cloud production rules in a knowledge-based system are modeled by CRPNs, where the truth degrees of places, the certainty factors of rules, and the thresholds of transitions are represented by interval clouds. Moreover, a matrix operation-based reasoning algorithm is proposed to improve the efficiency of calculating final truth degrees, in which both local and ordered weight coefficients are taken into consideration. Finally, a practical example concerning a power system is provided to demonstrate the usefulness and advantages of the proposed CRPN model.

Multiple observation improves the performance of 3D object classification. However, since the distribution of feature vectors obtained from multiple view points have strong nonlinear structure, the kernel-based methods are often introduced with nonlinear mapping. By mapping feature vectors to a higher dimensional space, kernel-based methods transform the distribution to weaken its nonlinearity. Although they have been succeeded in many applications, their computation cost is large. Therefore we aim to construct a comparable method with the kernel-based methods without using nonlinear mapping. Firstly we attempt to approximate a distribution of feature vectors with multiple local subspaces. Secondly we combine local subspace approximation with ensemble learning algorithm to form a new classifier. We will demonstrate that our method can achieve comparable performance with kernel methods through evaluation experiments using multiple view images of 3D objects from a public data set.