Abstract

Stable aerial manipulation during dynamic tasks such as object catching, perching, or contact with rigid surfaces necessarily requires compliant behavior, which is often achieved via impedance control. Successful manipulation depends on how effectively the impedance control can tackle the unavoidable coupling forces between the aerial vehicle and the manipulator. However, the existing impedance controllers for aerial manipulator either ignore these coupling forces (in partitioned system compliance methods) or require their precise knowledge (in complete system compliance methods). Unfortunately, such forces are very difficult to model, if at all possible. To solve this long-standing control challenge, we introduce an impedance controller for aerial manipulator which does not rely on a priori knowledge of the system dynamics and of the coupling forces. The impedance control design can address unknown coupling forces, along with system parametric uncertainties, via suitably designed adaptive laws. The closed-loop system stability is proved analytically and experimental results with a payload-catching scenario demonstrate significant improvements in overall stability and tracking over the state-of-the-art impedance controllers using either partitioned or complete system compliance.

Abstract

Do LLMs genuinely incorporate external definitions, or do they primarily rely on their parametric knowledge? To address these questions, we conduct controlleD experiments across multiple explanation benchmark datasets (general and domain-specific) and label definition conditions, including expert-curated, LLM- generated, perturbed, and swapped definitions. Our results reveal that while explicit label definitions can enhance accuracy and explainability, their integration into an LLM’s task-solvinG processes is neither guaranteed nor consistent, suggesting reliance on internalized representations in many cases. Models often default to their internal representations, particularly in general tasks, whereas domain-specific tasks benefit more from explicit definitions. These findings underscore the need for a deeper understanding of how LLMs process external knowledge alongside their pre-existing capabilities.

Abstract

Knowledge Graph Foundation Models (KGFMs) have shown promise in enabling zero-shot reasoning over unseen graphs by learning transferable patterns. However, most existing KGFMs rely solely on graph structure, overlooking the rich semantic signals encoded in textual attributes. We introduce SEMMA, a dual-module KGFM that systematically integrates transferable textual semantics alongside structure. SEMMA leverages Large Language Models (LLMs) to enrich relation identifiers, generating semantic embeddings that subsequently form a textual relation graph, which is fused with the structural component. Across 54 diverse KGs, SEMMA outperforms purely structural baselines like ULTRA in fully inductive link prediction. Crucially, we show that in more challenging generalization settings, where the test-time relation vocabulary is entirely unseen, structural methods collapse while SEMMA is 2x more effective. Our findings demonstrate that textual semantics are critical for generalization in settings where structure alone fails, highlighting the need for foundation models that unify structural and linguistic signals in knowledge reasoning.

Abstract

This paper presents an 18.87–24.19 GHz dual-core, dual-mode voltage-controlled oscillator (VCO) using a multi-tap (MT) inductor-based 2-port resonator with capacitive coupling via a mode-switching block. The 2-port resonator introduces a gate-to-drain phase shift, suppressing flicker phase noise (PN), while the high-quality factor of the tank minimizes thermal PN. The mode-switching block enables dual resonance modes—high-frequency band (HB) and low-frequency band (LB), achieving a wide frequency tuning range (FTR) and low PN in both the 1/f^2 and 1/f^3 regions. The proposed VCO is implemented in 65 nm CMOS, and post-layout results show a PN of –68.7 dBc/Hz at a 10 kHz offset and –118.2 dBc/Hz at a 1 MHz offset from an 18.87 GHz carrier frequency, with a power consumption of 25.65 mW and an area of 0.154 mm^2. The peak figure of merit (FoM) is 189.6 dBc/Hz, and the maximum 1/f^3 noise corner frequency is 347 kHz across the FTR.

Abstract

A conventional push–push frequency doubler (FD) is simple to implement but suffers from limited total efficiency ($eta_{text{total}}$) at millimeter-wave (mm-Wave) frequencies due to negative feedback introduced by the gate–drain capacitance ($C_{mathrm{gd}}$). To overcome this limitation, a 42.5-58 GHz mm-Wave FD is proposed, utilizing cross-coupled gate-to-source feedback. In the proposed topology, the push-push NMOS frequency doubler employs cross-coupled gate-to-source feedback, which generates a second-harmonic voltage at the gate that is in phase with the drain’s second-harmonic current. This feedback effectively mitigates the inherent feedback caused by $C_{mathrm{gd}}$, and doubles the gate-to-source voltage of the fundamental harmonic, $V_{gs}[f_{0}]$. As a result, the amplitude of the drain $2f_{0}$ current generated by the nonlinear second harmonic transconductance $g_{m2,nonlinear}$ of the NMOS increases, leading to enhanced drain efficiency ($eta_{mathrm{D}}$) and $eta_{text{total}}$. The proposed FD is designed in a 65 nm CMOS process. Post-layout simulations demonstrate a peak $eta_{mathrm{D}}$ of 28.39%, $eta_{text{total}}$ of 13.7%, and a peak $2f_0$ output power of 6.27 dBm at 47.5 GHz, with a 12 dBm local oscillator (LO) input and 14.92 mW DC power consumption. The doubler achieves a 3 dB bandwidth of 42.5–58 GHz $(30.85%)$ and an efficiency bandwidth 43.75–54.66 GHz $(20.4%)$ where $eta_{text{total}} geq 10%$.

Abstract

Respiratory diseases such as asthma, bronchitis, and pneumonia remain a leading cause of mortality worldwide, significantly affecting both individual health and socioeconomic conditions. Early and accurate diagnosis is essential to mitigate these effects, yet traditional diagnostic techniques such as auscultation are subjective and rely on the expertise of trained physicians. To address these challenges and improve diagnostic accessibility, this work proposes a fully portable and low-cost system for the automated detection and classification of respiratory diseases. Lung sounds are captured using a microphone interfaced with a microcontroller. These recordings are then transmitted to an Edge hardware platform, where a Convolutional Neural Network (CNN) model processes Mel spectrogram representations of the audio for classification. The system leverages the publicly available HF Lung dataset for training and evaluation. The proposed system achieved a classification accuracy of 80.5%, with a sensitivity of 95.65% and a specificity of 98.80%. This portable approach demonstrates the potential for real-time, AI-powered respiratory diagnostics in remote or resource-limited settings, bridging the gap between clinical expertise and everyday health monitoring.

Abstract

Frequency Modulated Continuous Wave (FMCW) radars, known for their millimeter-level accuracy, are increasingly used in healthcare for non-contact monitoring of vital signs like Heart Rate (HR) and Breath Rate (BR). HR and BR are estimated using frequency domain analysis of the extracted respiration and cardiac activity from Radar Data. The focus of our work is to extract the waveforms of chest vibrations due to respiration and cardiac activity, for which higher time resolution is required. This work utilizes a signal processing technique known as Zero Frequency Filtering (ZFF) on the FMCW Radar Data. ZFF is known to have the best time resolution. The processing pipeline is implemented on the PYNQ-ZU platform to ensure increased portability. Experimental results show that the proposed system achieves an accuracy of over 95.26% for HR estimation and over 98.47% for BR estimation, with a latency of less than 2 ms.

Abstract

We explore language-agnostic deep text embeddings for severity classification of dysarthria in Amyotrophic Lateral Sclerosis (ALS). Speech recordings are transcribed by human and ASR and embeddings of the transcripts are considered. Though speech recognition accuracy has been studied for grading dysarthria severity, no effort has yet been made to utilize text embeddings of the transcripts. We perform severity classification at different granularity (2, 3, and 5-class) using data obtained from 47 ALS subjects. Experiments with dense neural network based classifiers suggest that, though text features achieve nearly equal performances as baseline speech features, like statistics of mel frequency cepstral coefficients (MFCC), for 2-class classification, speech features outperform for higher number of classes. Concatenation of text embeddings and MFCC statistics attains the best performances with mean F1 scores of 88%, 68%, and 53%, respectively, in 2, 3, and 5-class classification.

Abstract

Grammar error detection in one's speech, referred to as spoken grammar error detection (SGED), is a critical component for building computer-assisted language learning systems. Traditionally, ASR followed by text-based GED has been used, but ASR errors can limit SGED performance. In this work, we analyse SGED under three speaking conditions: spontaneous, semi-spontaneous, and memorized, using five state-of-the-art ASRs. We examine the decoded text and ASR cues such as confidence scores and aligner probabilities, out of which, for the latter one, the ground-truth grammatically correct (GGC) text is used. Experiments revealed that autoregressive ASRs show bias toward GGC text, leading to suboptimal performance. It is observed that there is an increase in performance in response to the decreased freedom of speech, with semi-spontaneous speech outperforming memorized speech. Aligner probabilities outperform confidence scores despite the alignment and overconfidence issues.

Abstract

As robotic manipulators start performing more daily tasks, striking can be a useful method for transporting objects because it significantly increases the reachable workspace. However, striking methods are underexplored compared to pick-and-place because of the difficulty of modeling and executing striking interactions. In this paper, we develop an algorithm for striking objects so that they stop at a target location. We start with an optimizer in simulation that solves for the striking velocity given the relative target position, and perform system identification to set the simulation parameters. However, real-world striking with this model does not have high accuracy because it is unclear which parameters should be considered for identification in practice. Therefore, we finally developed a residual learning approach that subsumes all unmodeled differences between the simulation and the real-world environment into action, that is, striking velocity residuals. Our real-world experiments show that the residual learning model results in 81.6% more accurate object strikes.