Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain
the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in
Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles
and JavaScript.
Poly-reference system identification algorithms are mathematical tools to accurately estimate the dynamic properties of structural systems from vibration measurements. However, in some circumstance, they have a high computational cost and uncertainty. Amador and Brincker report an accurate and robust poly-reference formulation that can be easily implemented and demonstrate its performance by means of simulated and real-world application examples, including using a 15-storey heritage building.
A framework applicable in different imaging modalities can facilitate the medical imaging reconstruction efficiency but hindered by inefficient information communication across the data acquisition and imaging domains. Here, Jiadong Zhang and coworkers report a dual-domain generative framework to explore the underlying patterns across domains and apply their method to routine imaging modalities (computed tomography, positron emission tomography, magnetic resonance imaging) under one framework.
Xu and colleagues report cone beam computed tomography using a carbon nanotube X-ray source array. The approach addresses the limitations associated with conventional cone beam computed tomography using a single X-ray source, including image distortions, artefacts and low soft tissue contrast.
Innem Reddy and colleagues experimentally showcase data communications at speeds of up to 40 gigabits per second at terahertz frequencies with Bessel beams and validate their self-healing nature using a physical obstacle. The results suggest Bessel beams could be used for THz near-field communications, especially when mitigating the effects of blockages.
Johnson and Fiss successfully integrate a megawatt-scale latent heat storage system into a cogeneration thermal power plant to produce superheated steam. The data obtained demonstrates the feasibility of utilising latent heat for superheated steam in industry.
Nonlinear modulation within silicon microring modulators hinders the data transmission at optical interconnections. A neural network, proposed by Fangchen Hu, Yuguang Zhang, Hongguang Zhang and coworkers, can effectively mitigate the nonlinearity to accelerate the transmission speed. The results provide insights to integrate AI techniques to facilitate ultrahigh-speed silicon microring modulator design.
Carlos Rossa reports measurements exploring the impact of wind speed on the performance of photovoltaic modules. Data reveal that wind speed can increase the temperature dispersion in a module field, which can lead to unexpected losses. The findings could be used to optimise the performance prediction of photovoltaic fields.
Böhnert and colleagues demonstrate the weighted spin torque nano-oscillator, a spintronic circuit, as a basic programmable computing unit for neuromorphic computing systems. This circuit connects multiple spintronic devices with different functionalities in one circuit using a single fabrication process, which paves ways to fabricate more complex neuromorphic computing systems.
Microbubbles show great promise in clinical application as diagnostic ultrasound contrast agents. Chabouh and coworkers characterize the swimming motion of these microbubbles under cyclic external pressures through buckling. The results provide guidance to design ultrasound contrast agents with accurate mobility control.
Yu and Ouderji compare the thermodynamic performance of the flexible heat pump cycle with other performance enhancing cycles using a unified approach. Their analysis uncovers the working mechanisms and refrigerant characteristics that improve system performance, guiding heat pump optimisation.
Shaohua Ma and colleagues introduce a deep reinforcement based control framework for cable driven soft continuum robots with reduced computational time and high positioning accuracy. This work enables an easily implemented spatial partitioning technique for the control of cable-driven soft continuum robots for minimally invasive surgery.
Zhelyeznyakov and coworkers present a data-free physics-informed neural network to model and optimize the electromagnetic field distribution of large-scale ( ~ 1 mm in diameter) optical meta-lenses. This simplified method can speed up the design of large aperture meta-optics.
Existing indices available to quantify the influence of moisture sources on air humidity and temperature are limited in their ability to accurately describe dynamic phenomena. Jiale Hu and colleagues propose two dynamic indices and use them to experimentally evaluate the dynamic indoor environment change performance of a humidifier.
Tongyang Xu and Izzat Darwazeh use machine learning to identify a non-Nyquist waveform with high spectral efficiency, which is then practically validated with over-the-air image transmission using a hardware communication link. Waveforms such as these and the resulting spectral resource saving could be useful for enabling future 6G communications systems.
Mukherjee and colleagues develop a fully printed graphene-based capacitive sensor array for cognitive decision-making of robotic grippers. This rapid and low-cost solution endows determination of appropriate gripping location as well as detection of slippage and deformation of an object during manipulation.
Xin Xiong and colleagues estimate end-of life mass predictions of aerospace and rail vehicles in China and predict waste accumulation between 2000 and 2050. The study will aid in developing and managing effective technological solutions for a circular economy as well as formulating plans for governance.
Swarms of drones can collaborate to sense the environment. Nathan and coworkers propose a collective sensing strategy to improve tracking ability in densely forested areas, where targets can be occluded. The results pave the way to more accurate and sophisticated applications for detection of targets in complex environments.
Libo Tian, Jinbao Xia and colleagues introduce a gas sensor that can identify the components and concentrations of gases in a mixture. This method integrates optical sensing with deep learning, providing a solution for complex gas mixture analysis.
Ruibo Shang and coworkers introduce a Bayesian convolutional neural network (BCNN) to estimate the uncertainty of deep learning predictions in single-pixel imaging. The approach is a reliable tool where the confidence of image prediction needs to be approximated, such as simulated training data and hardware adjustments.
Shen and colleagues report a fingertip-type magnetic device for pulse detection and show that it works in conditions which might interfere with more conventional pulse detection stimuli, typically light or electric fields. The direct correspondence of the device signals to blood circulation induced vibrations hint also at potential utility in blood pressure measurements.