Search

As humans, our grasping and manipulation skills are highly dependent on our ability to perceive tactile properties. Conversely, most of today's robotic operations still rely predominantly on visual feedback for identifying the objects that need to be grasped and handled. In this work, we study the problem of recognizing everyday objects based solely on their tactile attributes. This has a significant practical value, as it could allow object identification even when visual sensing is impossible, or assist vision in difficult contexts. Our method consists of acquiring multi-modal tactile sensing data during a quick grasp-centric exploration phase, with minimal operational cost. Our algorithm was able to recognize objects from a considerably-large set of 50 general purpose items with an accuracy of 98.1%. Moreover, we show that it is possible to reliably identify a large proportion of these objects by only analyzing the deformation pattern that they undergo during compression. Further, we study our method's ability to learn relevant tactile properties to classify new objects. We also share our tactile sensing database that contains various sensor data acquired from more than 1600 experiments, which was used for this work. Finally, we discuss the relative performance and role of each tactile modality for differentiating objects.

We present a model for the growth kinetics of III-V nanowires which describes the time evolution of the monolayer coverage and group V concentration in the droplet. Special emphasis is put on the stopping effect at very low group V contents, where 2D island rapidly fills a fraction of monolayer and then grows much slower at the rate of the droplet refill from vapor. Such a regime has far-reaching implications in the periodically changing morphology of the growth interface, nucleation statistics, and size distributions within the ensembles of III-V nanowires.

This paper presents an Autonomous Underwater Vehicle (AUV) with a dynamic configuration of its actuation. The AUV is able to modify its configuration online (at each sampling time) during its missions. A procedure to optimize the dynamic configuration with respect to energy-like criterion is introduced. The simulation results are shown to prove the efficiency of a dynamic management of the actuation configuration.

In this paper, we summarize our hardware implementation for the DARPA Radio Frequency Machine Learning Systems (RFMLS) program. The system combines our wideband antenna array with past DARPA's investments in wideband technologies: BAE Systems' Hedgehog board with monolithic RF-to-baseband transceivers and Xilinx's RFSoC, a multi-processor, multi-converter FPGA. This combination of proven wideband technologies creates a powerful multi-channel software defined radio (SDR) controlled via an industry standard open interface. Measured active beam patterns of the digital array are also presented here to validate our hardware approach.

Increased arterial stiffness is one of the first signs of atherosclerosis. The objective of this study was to use non-invasive elastography (NIVE) to detect early changes in vascular biomechanics associated with obesity in children. The NIVE algorithm also measured the intimamedia thickness (IMT) for comparison.NIVE was applied in 120 children, 60 with elevated body mass index (BMI) (≥ 85th percentile for age and sex) and 60 non-overweight (BMI < 85th percentile). Participants were randomly selected from a longitudinal cohort, evaluating consequences of obesity in healthy children with one obese parent. The carotid wall was automatically segmented and elastograms were computed to measure the cumulated axial strain (CAS), cumulated axial translation (CAT), and maximal shear strain (Max |SSE|); IMT was also computed from segmented contours. Elastogram features were compared between groups with multivariate analyses to control for age, sex, Tanner stage, blood pressure, and lowdensity lipoprotein cholesterol (LDL).After Bonferroni correction, CAT was significantly higher in the elevated BMI group (0.68 ± 0.24 mm vs. 0.52 ± 0.18 mm), p < 0.001. CAS/CAT was significantly lower in the elevated BMI group (9.54 ± 4.8 %/mm vs. 13.34 ± 6.46 %/mm), p = 0.001; the lower CAS/CAT ratio suggests stiffer arteries with less deformation for a similar translation.Before Bonferroni correction, IMT was significantly higher in the elevated BMI group (0.36 ± 0.05 mm vs. 0.32 ± 0.05 mm), p = 0.013. IMT statistical difference was no longer significant after Bonferroni correction.After Bonferroni correction, NIVE detected differences in CAT and CAS/CAT biomarkers in elevated BMI children, whereas IMT failed to show a difference. NIVE is a promising technique to monitor radiological biomarkers of subclinical atherosclerosis in the pediatric population.

A new reconfigurable phased array antenna element is described. This element operates at X-band and provides digital control of the effective antenna structure, and therefore, its far field pattern, frequency response, polarization, etc. This presents an opportunity to improve the performance of software-defined or cognitive radio systems for which improved reconfigurable filtering is needed [1]–[2]. Because this reconfiguration occurs at the antenna and outside the front end, it has the potential to increase the dynamic range available to a system when operating in close proximity to other users.

This paper presents the properties and design procedure of the configuration matrix of over-actuated marine systems. Performance indices introduced in manipulator robots are extended in over-actuated marine vehicles. Moreover, two novel indices, namely reactive index and robust index, are proposed for configuration matrix design process. The problem is formulated as a multi-objective optimization problem. Simulation and preliminary experimental results show the solutions of the design process.

This paper presents an approach for the distributed modeling of a 4-port transistor. The proposed principle relies on considering any multi-finger transistor as the association of a number of elementary active linked together with extrinsic passive networks. An electromagnetic analysis allows to set the electrical equivalent scheme of the passive networks whereas the internal active device is defined by an equivalent model. It is shown how the equivalent intrinsic device (based on 2 fingers) and the values of the suitably defined distributed parasitic network elements can be accurately extracted and modeled on the basis of standard measurements. This approach is validated by the comparison of measured and simulated results for a GaAs HEMT transistor in the Ku-band.

The 2018 IROS Fan Robotic Challenge tasked participants with programming a robot to autonomously open and close a Spanish folding fan, highlighting the obstacles still associated with the dexterous manipulation of objects for robotic systems. Since high DoFs grippers are complex to coordinate and overkill for many industrial processes, our approach used an under-actuated parallel gripper with a 3D-printed adaptation to precisely grasp the fan in such a manner that gravity could be leveraged to act on the fan to produce an extrinsic, or external, dexterity. With our approach, we completed the challenge in 12.38 seconds, resulting in a top three finish. Furthermore, using a multi-modal tactile sensor, we analyzed the vibrations in the grasp during the manipulation and were able to distinguish the opening and closing of the fan from the motion of the robot with a 83% accuracy.

The Earth Networks Total Lightning Network (ENTLN) is comprised of wideband sensors to detect both the intra-cloud (IC) and cloud-to-ground (CG) flash signals efficiently. Since 2009, ENTLN has been used to track the properties of storms cells, such as the lightning flash rate, cell direction, and speed as a basis for issuing alerts for thunderstorms that could have the potential for severe weather. A storm cell is a cluster of flashes with a boundary as a polygon determined by the flash density value for a given period. From this storm cell data, Earth Networks produces the Dangerous Thunderstorm Alert (DTA). Most severe convective storms can generate high IC flash rate and high IC/CG flash-rate ratios. When rates exceed critical thresholds, DTAs are issued. DTA helps in predicting the storm ahead by generating an alert with significant lead time before the ground-level severe weather. Through DTA, the storm cells are monitored continuously, which enhances the situational awareness. In this study, we analyze the efficiency and accuracy of these automated DTAs by comparing with National Weather Service alert in relation to local storm reports, which are used as ground truth.

The need for disruptive advances in warfighter capability, upgradability, and utilization drives sensor systems toward convergent RF functionality with shared resources. The reliance of modern federated RF systems on redundant hardware leads to underutilization and inefficient use of SWaP-C. In pursuit of affordable solutions to this shortfall, converged RF and heterogeneous digital signal processing combine with existing open standard system-of-systems platform concepts to enable the next generation of platform and mission capability, through a true digital array interface to the RF environment.

Many aspects and courses that are part of the higher education system can be quite challenging for many students. Courses such as programming have been reported to be difficult by many students worldwide. In the modern age many courses that encompass new directions in science and professional work such as big data or data mining have emerged. These courses present new challenges for all students and are particularly challenging for students with disabilities, especially for visually impaired students. In this paper, the challenges of visually impaired students regarding higher education and visualization of data are presented and means and technologies for aiding visually impaired students in data visualization are discussed and elaborated. The research results about the usage of these technologies are also presented along with conclusions and comments about the possible future research directions in this area.

A new additive micro fabrication process for cavity resonator and band-pass filters is presented. 195 μm thick 3D air-filled components are fabricated operating at 38 GHz cavity resonator and two-pole band-pass filter. This paper presents a complete characterization of these components, with accurate simulation models. These models can be used to compute the losses in the component, and next improve them.

To estimate the detection efficiency of a ground based lightning location system, typically the ground based locations are compared to those seen by satellite. This works well because the detection efficiency of the satellite is relatively uniform with space, and reasonably good. However, this technique does not work for the 2015 or 2016 calendar years because the LIS satellite used is no longer in orbit, and has only recently been replaced by ISS-LIS. In this study, we will investigate applying a machine learning regression algorithm Lasso to lightning parameters determined by ENTLN to see if it is possible to estimate the detection efficiency of the network without satellite data.

This paper describes a new instantaneous wideband predistortion linearizer for TWTA enabling to use simultaneously broadband multi-carrier signals all over the complete 17.3–20.2GHz transmit Ka-band for the new generation of Telecom satellite. The linearizer has been embedded in a space equipment and successfully tested with a wideband Ka-band space TWTA. A state-of-the-art NPR performance was obtained over the entire 17.3–20.2GHz Ka-band with multi-carrier signals up to a 2.9GHz instantaneous bandwidth.

In this paper, we present an additive microfabrication process for the realization of millimeter-wave and near terahertz components. The fabrication of 3D air-filled components like air-filled cavity filters is possible with this technique. The demonstrated fabrication process can be used at Ka band.

This paper presents an extension of a three step de-embedding (Pad-Open-Short) method to a 3-port device for accurate on wafer MMIC S-parameters measurements. In the proposed method, an equivalent circuit-model using lumped elements is established according to the test-fixture. Furthermore, classical Pad-Open-Short method introduces systematic errors, observed beyond 20 GHz, due to perfect ‘Open’ and ‘Short’ standards assumption. This work also proposes a generalized Pad-Open-Short method with non-ideal standards. To validate the performance of this new method, reliable data were obtained from simulations and measurements of a GaAs transistor from UMS foundry operating up to 40 GHz.

We utilize rigorous coupled wave analysis (RCWA) to optimize the optical design of of GaAs nanowire solar cells. RCWA is highly accurate while being less computationally expensive than competing techniques. Using a simplex optimization method, we determine the optimal device geometries that maximize photocurrent density for bare nanowires and contacted nanowires in a dielectric material with a top ITO contact, finding agreement with previous work. We extend these results to include passivating shells, which are required for efficient devices. These optimizations give important guidance for geometric design of nanowire solar cell devices.

This paper describes a Volterra-based model of frequency dependent AM-AM and AM-PM conversions of a nonlinear traveling wave tube amplifier (TWTA) for use in communication system level simulation. The proposed method provides a systematic way to identify the model and ensures numerical stability in coefficient extraction. Comparison between simulation results and measurements are presented in terms of gain and noise power ratio (NPR).

One of the most common yet detrimental compression artifacts in frequency-domain audio codecs is known as pre-echo, which is perceived as a brief noise preceding transient signals, and is discernable even without direct comparison to the original signal. Because of its substantial negative impact on audio quality, many techniques have been proposed to alleviate it, but not without effect on coding efficiency. This paper presents a novel method to reduce pre-echo noise using only information already available at the decoder, such as scale-factors, that allow an estimation of the quantization noise levels in each frequency band. Doing so allows the proposed method to reduce pre-echo noise based on a precise modeling of the quantization noise spread before the transient signal. This has shown to improve both the subjective and objective quality of the MPEG AAC codec, and requires no modifications to the existent standard-compliant encoders.

The recent advent of Metagenome-Wide Association Studies (MGWAS) has allowed for increased accuracy in the prediction of patient phenotype (disease), but has also presented big data challenges. Meanwhile, Multiple Instance Learning (MIL) is useful in the domain of bioinformatics because, in addition to classifying patient phenotype, it can also identify individual parts of the microbiome that are indicative of that phenotype, leading to better understanding of the disease. We demonstrate a novel, efficient, and effective MIL-based computational pipeline to predict patient phenotype from MGWAS data. Specifically, we use a Bag of Words method, which has been shown to be one of the most effective and efficient MIL methods. This involves assembly of the metagenomic sequence data, clustering of the assembled contigs, extracting features from the contigs, and using an SVM classifier to predict patient labels and identify the most relevant read clusters. With the exception of the given labels for the patients, this entire process is de novo (unsupervised). We use data from a well-known MGWAS study of patients with Type-2 Diabetes and show that our pipeline significantly outperforms the classifier used in that paper, as well as other common MIL methods. We call our pipeline “CAMIL”, which stands for Clustering and Assembly with Multiple Instance Learning.

This paper presents a new high-speed and high-power switching circuit based on GaN HEMTs, well suited for envelope tracking supply modulators. The proposed switching cell operates up to 100MHz switching frequencies and provides output voltages up to 50V with high efficiency. This new topology takes advantage of a minimum number of components for reducing switching power losses. Moreover, this circuit acts like an inverting threshold comparator with rising and falling times in the order of one nanosecond. A theoretical analysis of static operation and a design approach are described and validated with measurements performed on a hybrid demonstration board.

International audience; Real long-term, complex and autonomous mission is still a challenge for robotics. This paper presents an efficient approach enhancing the robot with fault tolerance. It uses performance viewpoints to guide hardware and software resources allocation all along the mission according to faults effects and detection. Simulated and experimental results are proposed and analyzed.

We study vectorial bent functions with Dillon-type exponents. These functions have attracted attention because they are hyperbent whenever they are bent, and they achieve the highest possible algebraic degree among all bent functions on the same domain. In low dimensions we determine the simplest possible forms of such functions when they map to GF(4). We prove non-existence results for certain monomial and multinomial bent functions mapping to large codomains.