Search for: All records

Advancements in robotics and AI have increased the demand for interactive robots in healthcare and assistive applications. However, ensuring safe and effective physical human-robot interactions (pHRIs) remains challenging due to the complexities of human motor communication and intent recognition. Traditional physics-based models struggle to capture the dynamic nature of human force interactions, limiting robotic adaptability. To address these limitations, neural networks (NNs) have been explored for force-movement intention prediction. While multi-layer perceptron (MLP) networks show potential, they struggle with temporal dependencies and generalization. Long Short-Term Memory (LSTM) networks effectively model sequential dependencies, while Convolutional Neural Networks (CNNs) enhance spatial feature extraction from human force data. Building on these strengths, this study introduces a hybrid LSTM-CNN framework to improve force-movement intention prediction, increasing accuracy from 69% to 86% through effective denoising and advanced architectures. The combined CNN-LSTM network proved particularly effective in handling individualized force-velocity relationships and presents a generalizable model paving the way for more adaptive strategies in robot guidance. These findings highlight the importance of integrating spatial and temporal modeling to enhance robot precision, responsiveness, and human-robot collaboration. Index Terms —- Physical Human-Robot Interaction, Intention Detection, Machine Learning, Long-Short Term Memory (LSTM) 

In garnets from eclogites and blueschists formed within the subduction setting, fine-scale, oscillatory elemental zoning is a common feature, sometimes considered to record open-system fluid exchange during prograde metamorphism. We present oxygen isotope data for garnets with such zoning from five exhumed subduction zone complexes. Short length scale fluctuations in elemental and oxygen isotope zoning (which are themselves spatially decoupled) cannot be linked to open-system fluid exchange during garnet crystallization in all samples; these data do not provide evidence for a genetic relationship between elemental oscillations and fluid fluxing. However, garnets from one setting do provide clear evidence for syn-growth ingress of elementally and isotopically buffering fluids, a process that operated simultaneously with the formation of elemental oscillations. Our findings indicate multiple mechanisms of chemical transfer operate at the grain–rock scale during subduction, and that some subduction zone rocks may experience only limited interaction with external prograde fluids. These results are consistent with a picture of highly heterogenous volatile transfer during subduction, and suggest that some proportion of the fluid inventory inherited at shallow depths may be transferred to sub-arc depths. 

Orogenic ophiolites are a hallmark of Phanerozoic plate tectonics, containing igneous lithologies that provide constraints on fundamental tectono-magmatic processes. The c. 1900Ma Pembine Ophiolite (Wisconsin, USA) is associated with the Penokean Orogen and represents a rare example of a proposed Paleoproterozoic ophiolite. The Penokean Orogen shares broad characteristics with Phanerozoic (<541 Ma) orogens, but the origin of the Pembine Ophiolite remains unclear, with the mafic volcanic rocks interpreted as representing either an intra-oceanic arc or continental back arc setting. To test these hypotheses, we present the results of petrography, bulk-rock geochemistry and mineral chemistry for a suite of 34 Pembine rocks, as well as U-Pb zircon geochronology for two samples. Based on trace elements established as immobile in the studied rocks, we demonstrate that mafic volcanism progressed (up-stratigraphic-section) from mid-ocean ridge-like to boninitic. The chemical evolution is identical to that observed in < 250 Ma ophiolites in the Himalayan–Alpine Orogen, which record forearc spreading during the nascent stages of subduction in the Tethys Ocean. We interpret the Pembine Ophiolite as forearc lithosphere formed during subduction initiation and obducted to the margin of the Superior Craton during the Penokean Orogeny. The processes responsible for forming (and preserving) this example of a Paleoproterozoic ophiolite may not have been dissimilar to those operating on the Phanerozoic Earth. 

The Dadeville Complex of Alabama and Georgia (southeastern United States) represents the largest suite of exposed mafic-ultramafic rocks in the southern Appalachians. Due to poor preservation, chemical alteration, and tectonic reworking, a specific tectonic origin for the Dadeville Complex has been difficult to deduce. We obtained new whole-rock and mineral geochemistry coupled with zircon U-Pb geochronology to investigate the magmatic and metamorphic processes recorded by the Dadeville Complex, as well as the timing of these processes. Our data reveal an up-stratigraphic evolution in the geochemistry of the volcanic rocks, from forearc basalts to boninites. Our new U-Pb zircon crystallization data—obtained from three amphibolite samples—place the timing of forearc/protoarc volcanism no later than ca. 467 Ma. New thermobarometry suggests that the Dadeville Complex rocks subsequently experienced deep, high-grade metamorphism, at pressure-temperature conditions of ~7 kbar and ~760 °C. The data presented here support a model for formation of the Dadeville Complex in the forearc region of a subduction zone during subduction initiation and protoarc development, followed by deep burial/underthrusting of the complex during orogenesis.