Journal: Nature Machine Intelligence

Abbreviation

Nat Mach Intell

Publisher

Springer

Journal Volumes

ISSN

2522-5839

Description

Filters

20223
Reset filters

Search Results

Publications 1 - 3 of 3
  • Artificial microtubules for rapid and collective transport of magnetic microcargoes
    Item type: Journal Article
    Gu, Hongri; Hanedan, Emre; Boehler, Quentin; et al. (2022)
    Nature Machine Intelligence
    Directed transport of microcargoes is essential for living organisms as well as for applications in microrobotics, nanotechnology and biomedicine. Existing delivery technologies often suffer from low speeds, limited navigation control and dispersal by cardiovascular flows. In cell biology, these issues are largely overcome by cytoskeletal motors that carry vesicles along microtubule highways. Thus inspired, here we developed an artificial microtubule (AMT), a structured microfibre with embedded micromagnets that serve as stepping stones to guide particles rapidly through flow networks. Compared with established techniques, the microcargo travels an order of magnitude faster using the same driving frequency, and dispersal is mitigated by a strong dynamic anchoring effect. Even against strong fluid flows, the large local magnetic-field gradients enable both anchoring and guided propulsion. Finally, we show that AMTs can facilitate the self-assembly of microparticles into active-matter clusters, which then enhance their walking speed by bridging over stepping stones collectively. Hence, we demonstrate a unique strategy for robust delivery inside microvascular networks and for minimally invasive interventions, with non-equilibrium effects that could be equally relevant for enhancing biological transport processes.
  • A textile exomuscle that assists the shoulder during functional movements for everyday life
    Item type: Journal Article
    Georgarakis, Anna-Maria; Xiloyannis, Michele; Wolf, Peter; et al. (2022)
    Nature Machine Intelligence
    Effortlessly performing activities of daily living constitutes a cornerstone of our personal independence. Naturally, various forms of upper limb impairments can have a substantial impact on quality of life. We developed the Myoshirt, a textile-based soft wearable robot, or exomuscle, that autonomously follows the user’s movements and thereby assists the shoulder against gravity. With the Myoshirt, participants without impairments (n = 10, 5 male) experienced a delayed onset of muscular fatigue by 51.1 s (36.1%, P < 0.001), while during a functional task their muscular activity decreased by 49.1% (P < 0.001). Analogously, two participants with upper limb impairments due to a muscular dystrophy and a spinal cord injury experienced a delayed onset of muscular fatigue during unloaded arm lifts by 256.4 s (61.5%) and by 450.6 s (210.3%), respectively. Our evidence suggests that the Myoshirt is an effective tool that intuitively assists the shoulder during functional reaching tasks, with the potential of increasing the personal independence of people with upper limb impairments.
  • Enhanced spatio-temporal electric load forecasts using less data with active deep learning
    Item type: Journal Article
    Aryandoust, Arsam; Patt, Anthony; Pfenninger, Stefan (2022)
    Nature Machine Intelligence
    An effective way to mitigate climate change is to electrify most of our energy demand and supply the necessary electricity from renewable wind and solar power plants. Spatio-temporal predictions of electric load become increasingly important for planning this transition, while deep learning prediction models provide increasingly accurate predictions for it. The data that are used for training deep learning models, however, are usually collected at random using a passive learning approach. This naturally results in a large demand for data and associated costs for sensors such as smart meters, posing a large barrier for electric utilities when decarbonizing their grids. Here we investigate whether electric utilities can use active learning to collect a more informative subset of data by leveraging additional computation for better distributing smart meters. We predict ground-truth electric load profiles for single buildings using only remotely sensed data from aerial imagery of these buildings and meteorological conditions in the area of these buildings at different times. We find that active learning can enable 26–81% more accurate predictions using 29–46% less data at the price of 4–11 times more computation compared with passive learning.
Publications 1 - 3 of 3