Guilin Wen

Tensegrity structures, which combine rigidity and flexibility, can overcome the limitations of rigid robots in challenging terrains and the slow locomotion speed of soft robots. Their structural characteristics and mechanical properties are akin to musculoskeletal systems, offering valuable insights for the development of bioinspired robots. In this study, we present a galloping quadruped robot equipped with an innovative flexible tensegrity spine. The tensegrity spine successfully mimics the energy conversion mechanism of a natural spine. The elastic cables within the tensegrity spine store and release elastic potential energy during spinal flexion and extension. The released energy provides propulsion, enabling the robot to achieve higher speeds and leap over obstacles of a certain height. Additionally, the lateral bending and axial rotation of the spine enhance the robot's stability and maneuverability. We developed a Lagrangian dynamics model of the tensegrity spine, and dynamic analysis shows that cable stiffness directly regulates spinal morphology, energy efficiency, turning angle, and rotation angle. Additionally, rigid legs with unidirectional ratchet wheels are designed to replicate the mechanism of alternating ground contact between the front and hind legs. Extensive experiments show that the robot can gallop at a speed of 1.87 body lengths per second with optimized cable stiffness in the tensegrity spine, effectively navigate challenging terrains (such as grass, gravel roads, and steps), and successfully clear obstacles up to 0.5 body heights (100 mm) with its forelegs, while climbing slopes of up to 17.4°.

The creative combination of Gabriel horn (G-horn) hyperboloid with Finsler geometry opens a new door to black holes. G-horn with finite volume but infinite surface area can characterize the frame-dragging effects of strong gravitational fields. Finsler geometry is used to describe gravitational bounce depending on velocity direction. On the smooth and inextendible manifold representing by G-horn’s inner surface, the torsion determined by Cartan connection gives rise to centrifugal force and Coriolis force to counteract gravitational collapse. The Coriolis force serves as spin-flip trigger for gravitational bounce. The entering matters are choked at a certain inner-surface zone of horn’s neck as a novel non-Penrose’s trapped surfaces with global geodesic completeness, and forms a regular “hollow” core. Without any hidden or naked singularities, the mathematical singularity and the center of matter accumulating region need not coincide. By developing Cartan torsion-decomposing method, the competition mechanism between gravity of curvature and the “centrifugal force ± Coriolis force” of torsion is a elegant solution to those existing paradoxes or hypotheses such as black hole information loss, Penrose’s cosmic censorship, Weinberg’s asymptotically safe gravity. Further, the Schwarzschild, Kerr, Reissner‑Nordström black holes may be modified uniformly as this “singularity‑free” G-horn model.

Multiscroll chaotic systems have important applications in many engineering fields. Despite the growing research attention on the construction of multiscroll attractors, existing methods are afflicted by two critical limitations: insufficient generality and constrained direction arrangement of the multiple scrolls. This paper proposes a novel approach with exceptional performance regarding generality and diverse direction arrangement. The proposed approach is established based on generic autonomous dynamic systems and achieved through translation of the existing scrolls by predefined distances along any directions. Comparisons with other methods show that this approach enables more flexible control over the number, direction arrangement and spacing of scrolls. The application of this approach in information security is demonstrated by designing an encryption algorithm for image encryption and transmission. Performance analysis of this encryption algorithm reveals its high security. The excellent performance of this encryption algorithm implies a broader adaptability for practical applications of the proposed approach, which may inspire relevant research efforts.

Electroencephalography (EEG) is widely used in emotion recognition due to its reliability and noninvasiveness. However, challenges as signal noise and individual variability in EEG data present significant limitations for real-world applications. Inspired by recent neuroscience advances, we propose a contrastive twin latent informer (CTLIN) model for EEG-based emotion recognition and signal generation. The architecture employs a symmetric autoencoder (AE) that synergistically integrates PropSparse self-attention mechanisms with EEG feature extraction, incorporating a novel twin latent permutation (LP) design. By disentangling EEG features into subject and emotion latent representations through contrastive learning, the model demonstrates enhanced capability in capturing generalizable biomarkers. Experimental results show CTLIN’s state-of-the-art performance in zero-shot classification tasks. It significantly surpasses existing models in both subject identification and emotion classification across three benchmark datasets [SEED, SEED-IV, and a self-collected (SC) dataset], achieving subject classification accuracies of 98.63%, 99.13%, and 98.70%, respectively, with corresponding emotion classification accuracies of 77.92%, 53.79%, and 64.41%. Furthermore, a novel paradigm for EEG signal generation and data augmentation is established through the proposed LP approach, providing practical solutions to enhance emotion recognition.

This article proposes a long-term dynamic object relocalization (L-DOR) solution to address the challenge of mobile robots efficiently relocalize task-related objects in human–robot coexisting environments over extended periods. Existing methods mainly focus on one-time object localization, neglecting long-term relocalization amidst dynamic changes caused by human activities. To tackle this issue, a probabilistic model of object distribution based on spatio-temporal patterns is first established. Then, a robot-object perception interaction is introduced to achieve dynamic object category discrimination, enabling the robot to handle sporadic events caused by human activities. Besides, a cost-expectation balance-based object matching method is presented to determine the suitable task object and gradually infer its potential locations with probabilistic model. On this basis, we suggest a hierarchical global planning and local decision-making to prioritize search efforts to improve localization efficiency. Extensive comparisons and long-term experiments in real-world scenarios with a Fetch robot demonstrate the efficacy of L-DOR in terms of search performance, adaptability to different scenarios, and long-term effectiveness.