Search

In this work, we provide non-asymptotic, probabilistic guarantees for successful recovery of the common nonzero support of jointly sparse Gaussian sources in the multiple measurement vector (MMV) problem. The support recovery problem is formulated as the marginalized maximum likelihood (or type-II ML) estimation of the variance hyperparameters of a joint sparsity inducing Gaussian prior on the source signals. We derive conditions under which the resulting nonconvex constrained optimization perfectly recovers the nonzero support of a joint-sparse Gaussian source ensemble with arbitrarily high probability. The support error probability decays exponentially with the number of MMVs at a rate that depends on the smallest restricted singular value and the nonnegative null space property of the self Khatri-Rao product of the sensing matrix. Our analysis confirms that nonzero supports of size as high as $O(m^{2})$ are recoverable from $m$ measurements per sparse vector. Our derived sufficient conditions for support consistency of the proposed constrained type-II ML solution also guarantee the support consistency of any global solution of the multiple sparse Bayesian learning (M-SBL) optimization whose nonzero coefficients lie inside a bounded interval. For the case of noiseless measurements, we further show that a single MMV is sufficient for perfect recovery of the $k$ -sparse support by M-SBL, provided all subsets of $k + 1$ columns of the sensing matrix are linearly independent. [ABSTRACT FROM AUTHOR]

Pedestrian Detection is the most critical module of an Autonomous Driving system. Although a camera is commonly used for this purpose, its quality degrades severely in low-light night time driving scenarios. On the other hand, the quality of a thermal camera image remains unaffected in similar conditions. This paper proposes an end-to-end multimodal fusion model for pedestrian detection using RGB and thermal images. Its novel spatio-contextual deep network architecture is capable of exploiting the multimodal input efficiently. It consists of two distinct deformable ResNeXt-50 encoders for feature extraction from the two modalities. Fusion of these two encoded features takes place inside a multimodal feature embedding module (MuFEm) consisting of several groups of a pair of Graph Attention Network and a feature fusion unit. The output of the last feature fusion unit of MuFEm is subsequently passed to two CRFs for their spatial refinement. Further enhancement of the features is achieved by applying channel-wise attention and extraction of contextual information with the help of four RNNs traversing in four different directions. Finally, these feature maps are used by a single-stage decoder to generate the bounding box of each pedestrian and the score map. We have performed extensive experiments of the proposed framework on three publicly available multimodal pedestrian detection benchmark datasets, namely KAIST, CVC-14, and UTokyo. The results on each of them improved the respective state-of-the-art performance. A short video giving an overview of this work along with its qualitative results can be seen at https://youtu.be/FDJdSifuuCs. Our source code will be released upon publication of the paper. [ABSTRACT FROM AUTHOR]

Traditionally, road detection approaches mostly capitalize on RGB images, 3D LiDAR point cloud or their fusion. However, RGB camera is sensitive to light conditions, while LiDAR point cloud is sparse compared with dense image pixels. In this work, a new hybrid image dataset is provided for the task of road detection based on cameras. In this dataset, the hybrid images are acquired by an optically aligned hybrid imaging device, consisting of a far-infrared (FIR) imager and an RGB camera to output pixel-wise registration of thermal and RGB frames. Then we investigate on three methods based on fully convolutional neural network (F-CNN) to demonstrate the advantages by fusing RGB-FIR images in road detection. First, a middle-fusion based model is built, where the output feature maps of encoder branches from RGB and FIR images are directly concatenated into a single-fusion branch as the decoder. Next, the originally discarded layers after fusion operation for both RGB and FIR branches are recovered as the mimic branches to imitate the distributions of the fusion outputs, which constitutes an extended cross model (ECM). Moreover, the outputs of mimic branches at different scales are also used to imitate the corresponding outputs in the fusion branch, called a hierarchical cross model (HCM). The experimental results demonstrate the effectiveness and efficiency of our fusion strategies. [ABSTRACT FROM AUTHOR]

Electromyography (EMG) is one of the most common methods to detect muscle activities and intentions. However, it has been difficult to estimate accurate hand motions represented by the finger joint angles using EMG signals. We propose an encoder-decoder network with an attention mechanism, an explainable deep learning model that estimates 14 finger joint angles from forearm EMG signals. This study demonstrates that the model trained by the single-finger motion data can be generalized to estimate complex motions of random fingers. The color map result of the after-training attention matrix shows that the proposed attention algorithm enables the model to learn the nonlinear relationship between the EMG signals and the finger joint angles, which is explainable. The highly activated entries in the color map of the attention matrix derived from model training are consistent with the experimental observations in which certain EMG sensors are highly activated when a particular finger moves. In summary, this study proposes an explainable deep learning model that estimates finger joint angles based on EMG signals of the forearm using the attention mechanism. [ABSTRACT FROM AUTHOR]

With the expansion of sensor nodes to newer avenues of technologies, such as the Internet of things (IoT), internet of bodies (IoB), augmented reality (AR), and mixed reality, the demand to support high-speed operations, such as audio and video, with a minimal increase in power consumption is gaining much traction. In this work, we focus on these nodes operating in audio-based AR (AAR) and explore the opportunity of supporting audio at a low power budget. For sensor nodes, communicating one bit of data usually consumes significantly higher power than the power associated with sensing and processing/computing one data bit. Compressing the number of communication bits at the expense of a few computation cycles considerably reduces the overall power consumption of the nodes. Audio codecs such as AAC and LDAC that currently perform compression and decompression of audio streams burn significant power and create a floor to the minimum power possible in these applications. Compressive sensing (CS), a powerful mathematical tool for compression, is often used in physiological signal sensing, such as EEG and ECG, and it can offer a promising low-power alternative to audio codecs. We introduce a new paradigm of using the CS-based approach to realize audio compression that can function as a new independent technique or augment the existing codecs for a higher level of compression. This work, CS-Audio, fabricated in TSMC 65-nm CMOS technology, presents the first CS-based compression, equipped with an ON-chip DWT sparsifier for non-sparse audio signals. The CS design, realized in a pipelined architecture, achieves high data rates and enables a wake-up implementation to bypass computation for insignificant input samples, reducing the power consumption of the hardware. The measurement results demonstrate a 3X–15X reduction in transmitted audio data without a perceivable degradation of audio quality, as indicated by the perceptual evaluation of audio quality mean opinion score (PEAQ MOS) ${>}1.5$. The hardware consumes 238 $~\mu \text{W}$ power at 0.65 V and 15 Mbps, which is ($\sim $ 20X–40X) lower than audio codecs. [ABSTRACT FROM AUTHOR]

Besides the large amount of information multispectral imaging offers, multispectral visual odometry remains overlooked due to the dissimilarity between modalities. In order to tackle the challenging feature matching between multispectral stereo images and to overcome the lack of robust multispectral visual localisation solutions, a novel approach is proposed in this paper. It consists in tracking features in each modality simultaneously, in a monocular manner and then, estimating motion in a windowed bundle adjustment framework and using the geometry of the stereo setup to recover the missing scale. The estimation is robustified by selecting adequate keyframes based on feature parallax and by maximising the mutual information between all the features reprojected in the stereo pair. Furthermore, the proposed multispectral visual odometry solution is integrated in an error-state Kalman filter framework to deal with challenging environments, where the quality of images is reduced. Two measurements models, using absolute and relative camera poses, are presented. The superiority of relative poses is then shown by providing a failure recovery algorithm which relies on inertial data when visual data are not accessible. The algorithm was tested on innovative series of visible-thermal multispectral datasets, acquired from a car with real driving conditions. An overall error of around 2% of the travelled distance was achieved on these datasets. [ABSTRACT FROM AUTHOR]

On-demand data sensing and wireless power transfer (WPT) can provide sustainability and robust operations in large-scale industrial Internet of Things (IoT) networks. The efficiency of on-demand data collection and WPT can be increased by efficient scheduling of IoT nodes and dedicated energy transmitters respectively. In this article, we propose an energy-aware mode switching strategy to enable IoT nodes to perform either on-demand sensing or dedicated WPT. For on-demand sensing, we propose an IoT node scheduling scheme to maximize the utility of the IoT nodes comprising residual energy and energy required for sensing operation while considering the reliability of sensing tasks. For WPT, we propose an energy transmitter scheduling scheme for IoT nodes to minimize the cost of charging while keeping IoT nodes sufficiently charged. The simulation results for IoT node scheduling demonstrate that less than 50% IoT nodes need to be activated in all scenarios to complete the tasks. The proposed energy transmitter scheduling scheme shows that less than 60% energy transmitters should be scheduled in all the scenarios which results in significant energy reduction in the overall system. [ABSTRACT FROM AUTHOR]

Manufacturing of microsystems based on complex three-dimensional microstructures requires critical steps such as mass calibrations and integration process. However, such steps involve several manipulation steps which are not well controlled thus the fragile 3D microstructures could be damaged or destroyed. In this work, we have demonstrated an AFM based direct mass measurement and manufacturing of 3D helical microswimmer. The proposed method shows to be rapid, repeatable and minimally-destructive thanks to the precise force control and manipulation by AFM. The proposed micromanipulation steps consist of the pick-measure-integrate manipulation steps using van der Waals force-based attachment and the mass measurement by resonant frequency shift. For the testing structures, we fabricated the 6 different types of 3D helical microswimmers vertically fabricated on the conical shape microneedle supports for uniform surface metallization and facile detachment. With the mass measurement sensitivity of 25 Hz/pg and the direct integration to microfluidics, we successfully demonstrated the 3D propulsion and non-contact micromanipulation by 3D helical microswimmer in microfluidics. [ABSTRACT FROM AUTHOR]

Drastic reduction in the manufacturing cost of sensors and actuators has resulted in considerable growth in the number of smart objects. The so-called Internet of Things (IoT) blends the real and virtual environments and removes time and distance barriers. It is widely perceived as a major enabler for the efficient and effective provision of services across a range of sectors. It has naturally attracted the interest of cyberattackers. Due to the heterogeneity, resource-constraints, scale, and internet connectivity of IoT devices, each IoT layer is prone to various threats. Intruders consider the network layer of IoT as the gateway and leverage vulnerabilities in the routing protocol to compromise the Confidentiality, Integrity, and Availability (CIA) of connected nodes. Researchers have proposed different security infrastructures to mitigate harm to IoT networks. One of these is the Intrusion Detection System (IDS). An IDS is an essential component for the network security layer and is widely adopted to reinforce the security of the IoT network. This systematic literature review explores the IPv6 Routing Protocol for Low Power and Lossy Networks (RPL) and its existing threats, classifies relevant IDS techniques and identifies areas requiring further investigation. We review 103 published papers in this domain. [ABSTRACT FROM AUTHOR]

3D optical sensors are becoming more and more popular in vision guidance of industrial robots and other industrial automation applications. Although research on high-speed 3D shape measurement (or imaging) has experienced tremendous growth over the past decades, simultaneously achieving high-speed and high-accuracy performance remains a major challenge in industrial practice. This paper presents a new FPGA architecture for the PMP algorithm and designs an high-speed embedded binocular structured light 3D measurement sensor. The sensor mainly contains a DLP projector, two cameras and a Xilinx Zynq-7000 UltraScale which make real-time computation possible. Experiments showed the time consuming of processing a set of 9 × 2 images with a resolution of 2048 × 1536 using the sensor proposed in this paper is 31.47 ms that has a better performance than 255ms on GPU. To the best of our knowledge, this is the first high-speed FPGA-based embedded binocular structured light 3D measurement sensor. This proposed 3D sensor can obtain ultra high quality 3D information in real time, it can be widely applied to robot random pin-picking and other industrial applications. [ABSTRACT FROM AUTHOR]

This article reviews the investigation of the mode-localized microelectromechanical system (MEMS) coupled resonant sensors. This class of a sensor features a new approach in the sensing method (i.e. sensing shifts in the eigenstates/amplitude ratio (AR) instead of a frequency shift) in resonant devices. The key performance metrics and progress in terms of sensitivity, noise optimization, resolution, output stability, and bandwidth are closely observed for the resonant devices that use multi-degree of freedom (m-DoF) resonators for transduction and subsequent sensing. A critical analysis, insight, and wide-ranging applications are studied for the mode-localized sensors that operate in a vacuum (with a higher quality factor, ${Q}$) or in an atmospheric pressure environment. Parameters such as the role of the operating/bias point, modal overlap, different forms of the output, and nonlinearity in the output are analyzed. The impact of structural asymmetry and how it can contribute to the performance enhancement of the sensor design is reported. An operation of the mode-localized sensors in the open/closed-loop configuration, in linear and nonlinear regions, is presented to understand the attempts made to enhance/optimize the sensor performance. Several potential applications developed particularly in the past five-year time span are discussed. Future endeavors and concluding remarks are offered for this emerging class of a coupled resonant (CR) sensor that features unprecedented ultrahigh parametric sensitivity, parallel detection capability of multiple analyte/s, and inherent common-mode rejection. [ABSTRACT FROM AUTHOR]

Recent advancements in perception for autonomous driving are driven by deep learning. In order to achieve robust and accurate scene understanding, autonomous vehicles are usually equipped with different sensors (e.g. cameras, LiDARs, Radars), and multiple sensing modalities can be fused to exploit their complementary properties. In this context, many methods have been proposed for deep multi-modal perception problems. However, there is no general guideline for network architecture design, and questions of “what to fuse”, “when to fuse”, and “how to fuse” remain open. This review paper attempts to systematically summarize methodologies and discuss challenges for deep multi-modal object detection and semantic segmentation in autonomous driving. To this end, we first provide an overview of on-board sensors on test vehicles, open datasets, and background information for object detection and semantic segmentation in autonomous driving research. We then summarize the fusion methodologies and discuss challenges and open questions. In the appendix, we provide tables that summarize topics and methods. We also provide an interactive online platform to navigate each reference: https://boschresearch.github.io/multimodalperception/. [ABSTRACT FROM AUTHOR]

Assistive vision-based solutions for the driver extend the capabilities of human vision and support safe travel. Unfortunately, their widespread usage is generally limited to expensive cars. Interestingly, a high price is most likely a derivative of the costs incurred in the research instead of the value of hardware components. In the article we show that a mobile system for pedestrian detection in severe lighting conditions can be build using state of the art algorithms and widely available hardware. The proposed night-vision system for pedestrian detection processes thermal images using a proprietary ODROID XU4 microcomputer under Ubuntu MATE operating system. We applied a cascade object detector for the task of human silhouette detection in context of thermal imagery and contrasted the results with the state of the art deep learning approach. The experiments conducted prove effectiveness of the proposed solution. [ABSTRACT FROM AUTHOR]

In the Internet of Things networks deploying fixed infrastructure is not always the best and most economical solution. Advances in efficiency and durability of Unmanned Aerial Vehicles (UAV) made flying small cell base stations (BS) a promising approach by providing coverage and capacity in environments where using fixed infrastructure is not economically justified. A key challenge in covering an area with UAV-based small cell BSs is optimal positioning the UAVs to maximize the coverage and minimize the number of required UAVs. In this paper, we propose an optimization problem that helps to determine the number and position of the UAVs. Moreover, to have efficient results in a reasonable time, we propose complementary heuristic methods that effectively reduce the search space. The simulation results show that our proposed method performs better than genetic algorithms. [ABSTRACT FROM AUTHOR]

We present a hybrid multiplexing (HYM) scheme for MRAMs with separate read-write paths, employing different bit-cell selection methods during read and write based on their suitability for data sensing and switching operations. The proposed technique allows the sharing of the read access transistor among the cells belonging to the same word, leading to area savings and enhancement in distinguishability. Furthermore, by employing dummy bits in an array and setting their values according to the process corner, the spread of the read current due to global variations can be tightened. We apply our technique on MRAM with spin Hall effect-based write and show that the HYM achieves 20% to 25% area reduction along with 2.4X improvement in cell TMR, 25% to 38% lower write power, and upto 50% increase in the read disturb margin compared to the standard multiplexing. The spread of the read current reduces by 38% to 53% by employing the dummy bits. [ABSTRACT FROM PUBLISHER]

Spin-transfer torque random access memory (STT-RAM) has demonstrated great potentials as a universal memory for its fast access speed, zero standby power, excellent scalability, and simplicity of cell structure. However, large process variations of both magnetic tunneling junction (MTJ) and CMOS process severely limit the yield of STT-RAM chips and prevent the massive production from happening. In this paper, we analyze and compare the impacts of process variations on various sensing schemes of STT-RAM design. On top of it, we propose a novel voltage-driven nondestructive self-reference sensing scheme to enhance the STT-RAM chip yield by significantly improving sense margin. Monte Carlo simulations of a 16-Kb STT-RAM array shows that our proposed scheme can achieve the same yield as the previous nondestructive self-reference sensing scheme while improving the sense margin by five times with the similar access performance and power. [ABSTRACT FROM AUTHOR]

The Pd/n-\SnOx/i-diamond/p-diamond diodes prepared by field-enhanced hot-wire CVD (HWCVD) (FEHWCVD) system on a silicon substrate with nanotip structures are studied systematically. Both the nanotip structure and the better film quality deposited by the FEHWCVD lead the developed p-i-n diamond diode to have a high relative sensitivity ratio of \sim91% to 100-ppm-carbon monoxide (CO)-gas ambient. The sensitivity ratio is better than 79% compared to the one without the nanotip and prepared by the conventional HWCVD. Thus, the developed p-i-n diamond diode has better potential for high-temperature CO sensing applications. [ABSTRACT FROM PUBLISHER]

Among all the emerging memories, Spin-Transfer Torque Random Access Memory (STT-RAM) has demonstrated many promising features such as fast access speed, nonvolatility, excellent scalability, and compatibility to CMOS process. However, the large process variations of both magnetic tunneling junction (MTJ) and MOS transistors in the scaled technologies severely limit the yield of STT-RAM chips. In this work, we proposed a new sensing scheme, named as nondestructive self-reference sensing, or NSRS, for STT-RAM. By leveraging the different dependencies of the high and low resistance states of MTJs on the cell current amplitude, the proposed NSRS technique can work well at the scenario when bit-to-bit variation of MTJ resistances is large. Furthermore, we proposed three combined magnetic- and circuit-level techniques, including R-I curve skewing, yield-driven cell current selection, and ratio matching, to further improve the sense margin and robustness of NSRS sensing scheme. The measurement results of a 16 Kb STT-RAM test chip show that our proposed nondestructive self-reference sensing technique can reliably readout all the measured memory bits, of which 10% read failure rate was observed by using the conventional sensing technique. The three enhancement technologies ensure a 20 mV minimum sense margin and the whole sensing process can complete within 15 ns. [ABSTRACT FROM PUBLISHER]

In this letter, a new method of improving the quality factor of microresonators inside liquids is presented. By utilizing a conformally coated thin layer of parylene, which is highly hydrophobic, we trap air between the narrow gaps of a lateral resonator and prevent liquids from entering the gaps, thus preventing the viscous damping increase on the side surfaces. Initial tests with capacitive comb drives show that the quality factor decreases only threefold (from 157 to 55) when the device is operated in water after parylene coating. This method is very promising for in-liquid resonator applications, particularly real-time bioagent detection where high quality factor is essential for high mass resolution.\hfill[2011-0110] [ABSTRACT FROM AUTHOR]