Advanced intelligent systems (Weinheim an der Bergstrasse, Germany)最新文献

Neuromorphic devices, inspired by the human brain's efficiency and adaptability, hold great potential for artificial intelligence (AI) hardware to overcome the limitations of traditional von Neumann architecture. As a subclass, multimodal and multifunctional neuromorphic devices have recently gained a lot of attention due to their advantages in in-sensor computing and sophisticated behaviors. In this review, recent advances in materials, device structures, and applications in this field are systematically presented. It includes optical, electrical, mechanical, and chemical sensing in multimodal neuromorphic device, which enable in-sensor computing to minimize energy consumption and enhance real-time decision-making. The materials applied in this field such as phase-change, 2D materials, and ferroelectrics are summarized for their roles in achieving synaptic plasticity, nonvolatile memory for multifunctional neuromorphic devices. Structural innovations, including reconfigurable, multi-terminal, and 3D-integrated designs, further optimize parallel processing and multifunctional integration. Besides, application scenarios of multimodal and multifunctional neuromorphic devices and their advantages for improving the efficiency of AI are reviewed. Finally, challenges in material stability and commercialization are discussed, it emphasizes the need for interdisciplinary efforts to bridge the gap. This review provides critical insights and future directions for developing brain-inspired, energy-efficient AI hardware.

Non-small cell lung cancer (NSCLC) comprises the largest subtype of lung cancer with the most cases. Lung adenocarcinoma and lung squamous cell carcinoma are two NSCLC subtypes that pose challenges for accurate diagnosis using conventional methods, including histological examination and imaging, which can be slow and inconclusive. To address these concerns, RPSLearner is proposed, which combines random projection (RP) for dimensionality reduction and stacking ensemble learning to accurately predict lung cancer subtypes. Specifically, multiple independent RP matrices are first generated to project the high-dimensional RNA-seq data into a lower-dimensional space, whose features are subsequently concatenated. After that, the concatenated RP features are fed into a stack of diverse base classifiers, and integrated the predictions from base models via a deep linear layer network. Benchmarking tests on 1 333 NSCLC patients demonstrated that RPSLearner outperformed state-of-the-art approaches for lung cancer subtype classification. Specifically, RPSLearner efficiently preserved sample-to-sample distances even after significant dimension reduction, and the meta-model in RPSLearner yielded consistently higher scores than individual base models. In addition, the feature fusion method outperformed conventional score ensemble methods. We believe RPSLearner is a promising model for downstream lung cancer clinical diagnosis, and it holds the potential to be extended to subtyping of other types of cancer.

The postoperative rehabilitation of ankle fractures, particularly in the home setting, has a crucial influence on the recovery of lower limb function. To enhance the portability, real-time performance, and safety of postoperative remote rehabilitation training, this study proposes a novel robot-assisted remote rehabilitation system tailored for postoperative ankle fracture patients. Based on a distributed system architecture, the hardware system enables modular decomposition and facilitates wireless control of the lower controller. The total weight of the robotic system is 2.634 kg. By combining a deep learning algorithm with an interpolation fitting method, the time delay in interaction force signals during remote communication is predicted and compensated. The control frequency is elevated to 100 Hz with a maximum normalized root mean square error of 10.89%, ensuring the precision and continuity of the robot control system. Additionally, a full-cycle rehabilitation training strategy based on adaptive admittance control with system stiffness identification is proposed, encompassing passive, active–passive, isotonic, and active activities of daily living trainings. Experimental results indicate that the robotic system can execute the training strategies at each phase with high accuracy and safety, and the proposed adaptive control strategy has better compliance than fixed parameter admittance control and fuzzy admittance control methods.

Tensegrity Robotics

This research presents tensegrity articulated joints with actuation that combine thin pneumatic artificial muscles and energy-restoring elastics, both integrated into the tensile network. It uses a tensegrity spine-inspired topology, further refined through a multi-objective, constraint-based evolutionary algorithm. The method was validated by designing and fabricating two types of joints, which were tested in a quadruped robot and gripper application. More details can be found in the Research Article by Jan Petrš and co-workers (Doi: 10.1002/aisy.202500310).

Cost of Morphing

Adaptive morphogenesis enables robots to navigate diverse environments with enhanced efficiency. JART, an amphibious quadruped, uses laminar jamming with kirigami-cut layers to switch between load-bearing and hydrodynamic limb configurations. This approach reduces morphing energy by 98.5% compared to thermally driven systems, without compromising terrestrial or aquatic performance. The design advances energy-efficient, shape-morphing robotics for multidomain locomotion. More details can be found in the Research Article by Rebecca Kramer-Bottiglio and co-workers (Doi: 10.1002/aisy.202401055).

The remarkable capability of living organisms to reshape themselves—whether the compliant deformation of musculoskeletal structure, the origami-like folding of a butterfly's wing, or the stiffness modulation of a turtle flipper—continues to inspire breakthroughs in robotic design. Nature reveals that form and function are inherently coupled and adaptability is rooted in intelligent mechanical design, e.g., materials and structures. This special issue of Advanced Intelligent Systems assembles the latest progress in shape-morphing robotics and machines that alter their physical geometry to achieve adaptive properties in response to changing environments, task demands, or unexpected disturbances.

Shape-morphing robots lie at the intersection of materials science, mechanical design, and intelligent control. Several contributions translate biological strategies directly into engineered systems. Khan et al. (doi: 10.1002/aisy.202400620) fabricate 3D-printed magnetic butterflies whose vein-embedded composites replicate monarch wing folding, delivering agile, energy-efficient aerial maneuvers. Ramirez et al. (doi: 10.1002/aisy.202401055) present JART, an amphibious robot that uses kirigami-layer jamming to toggle between flippers and legs, cutting morphing energy by approximately 98% relative to thermally driven designs. Jensen et al. (doi: 10.1002/aisy.202500422) describe DAWN, a dual-helical, wave-propelled crawler whose elastomer skins boost friction and shield internal linkages, letting it traverse sand, gravel, and wet soil. Huang et al. (doi: 10.1002/aisy.202500365) mimic human digits in a wearable pneumatic physiotherapy device that integrates actuation, sensing, and control for on-body, customizable massage. These studies showcase how biomimicry delivers both functional versatility and structural elegance, yet also highlight enduring challenges—continuous materials integration, long-term durability, and tightly coupled actuation–sensing architectures.

A recurring theme in this issue is multi-modal morphologies: the robots will tailor not only shape, but also properties, such as stiffness, for high load-bearing capabilities. In terms of shapes, Chen et al. (doi: 10.1002/aisy.202500123) achieve this through tunable-stiffness architecture morphing structures (TSAMS) that exhibit over 300-fold tunable stiffness range using shape memory alloy actuators. Samarakoon et al. (doi: 10.1002/aisy.202400417) further our understanding of the tiling robots through design, kinematic modeling and control of polyform-inspired robots capable of transforming between two polymorphic shapes, and establishing a taxonomy for scalable morphing tiling robots. Petrš et al. (doi: 10.1002/aisy.202500310) integrate McKibben muscles and elastic cords across tensegrity structures to create distributed, fault-tolerant morphing joints with variable stiffness.