Audio-visual speech recognition (AVSR) system is thought to be one of the most promising solutions for reliable speech recognition, particularly when the audio is corrupted by noise. However, cautious selection of sensory features is crucial for attaining high recognition performance. In the machine-learning community, deep learning approaches have recently attracted increasing attention because deep neural networks can effectively extract robust latent features that enable various recognition algorithms to demonstrate revolutionary generalization capabilities under diverse application conditions. This study introduces a connectionist-hidden Markov model (HMM) system for noise-robust AVSR. First, a deep denoising autoencoder is utilized for acquiring noise-robust audio features. By preparing the training data for the network with pairs of consecutive multiple steps of deteriorated audio features and the corresponding clean features, the network is trained to output denoised audio features from the corresponding features deteriorated by noise. Second, a convolutional neural network (CNN) is utilized to extract visual features from raw mouth area images. By preparing the training data for the CNN as pairs of raw images and the corresponding phoneme label outputs, the network is trained to predict phoneme labels from the corresponding mouth area input images. Finally, a multi-stream HMM (MSHMM) is applied for integrating the acquired audio and visual HMMs independently trained with the respective features. By comparing the cases when normal and denoised mel-frequency cepstral coefficients (MFCCs) are utilized as audio features to the HMM, our unimodal isolated word recognition results demonstrate that approximately 65 % word recognition rate gain is attained with denoised MFCCs under 10 dB signal-to-noise-ratio (SNR) for the audio signal input. Moreover, our multimodal isolated word recognition results utilizing MSHMM with denoised MFCCs and acquired visual features demonstrate that an additional word recognition rate gain is attained for the SNR conditions below 10 dB.

/pdf/audio-visual-speech-recognition-using-deep-learning-2nrljpc31c.pdf

Audio-visual speech recognition using deep learning

Needle insertion is a critical aspect of many medical treatments, diagnostic methods, and scientific studies, and is considered to be one of the simplest and most minimally invasive medical procedures. Robot-assisted needle steering has the potential to improve the effectiveness of existing medical procedures and enable new ones by allowing increased accuracy through more dexterous control of the needle-tip path and acquisition of targets not accessible by straight-line trajectories. In this article, we describe a robot-assisted needle-steering system that uses three integrated controllers: a motion planner concerned with guiding the needle around obstacles to a target in a desired plane, a planar controller that maintains the needle in the desired plane, and a torsion compensator that controls the needle-tip orientation about the axis of the needle shaft.

/pdf/robot-assisted-needle-steering-spqto96cex.pdf

Robot-Assisted Needle Steering

Elastomeric proteins act as the essential functional units in a wide variety of biomechanical machinery and serve as the basic building blocks for biological materials that exhibit superb mechanical properties. These proteins provide the desired elasticity, mechanical strength, resilience, and toughness within these materials. Understanding the mechanical properties of elastomeric protein-based biomaterials is a multiscale problem spanning from the atomistic/molecular level to the macroscopic level. Uncovering the design principles of individual elastomeric building blocks is critical both for the scientific understanding of multiscale mechanics of biomaterials and for the rational engineering of novel biomaterials with desirable mechanical properties. The development of single-molecule force spectroscopy techniques has provided methods for characterizing mechanical properties of elastomeric proteins one molecule at a time. Single-molecule atomic force microscopy (AFM) is uniquely suited to this purpose. ...

/pdf/protein-mechanics-from-single-molecules-to-functional-e63loesb7o.pdf

Protein Mechanics: From Single Molecules to Functional Biomaterials

Fused deposition modelling is an additive manufacturing technology that is widely employed to produce good quality products with complex geometries at a low cost and with efficient manufacturing and delivery logistics. The mechanical properties can be enhanced by studying the numerous FDM parameters and by using new materials. In this work, was studied the mechanical properties tensile strength, flexural strength, and impact energy of 3D printed parts manufactured with FDM technology and PLA-graphene raw material by varying the infill and layer thickness parameters using a statistical technique CCD—central composite design. Due to the layered production process, 3D printed parts exhibit anisotropic behaviour. In the tests, the flat orientation and honeycomb infill pattern were maintained. The results showed that the mechanical properties improve as the linear layer thickness parameter increases. The behaviour was different in each test for the linear infill parameter. The mechanical properties, tensile strength and flexural strength, increased as the infill increased, while impact energy decreased as infill increased. The relationship between mechanical properties and printing time/weight was also evaluated.

Mechanical properties of PLA-graphene filament for FDM 3D printing

In this paper, we propose an energy management strategy based on deep reinforcement learning for a hybrid battery system in electric vehicles consisting of a high-energy and a high-power battery pack. The energy management strategy of the hybrid battery system was developed based on the electrical and thermal characterization of the battery cells, aiming at minimizing the energy loss and increasing both the electrical and thermal safety level of the whole system. Primarily, we designed a novel reward term to explore the optimal operating range of the high-power pack without imposing a rigid constraint of state of charge. Furthermore, various load profiles were randomly combined to train the deep Q-learning model, which avoided the overfitting problem. The training and validation results showed both the effectiveness and reliability of the proposed strategy in loss reduction and safety enhancement. The proposed energy management strategy has demonstrated its superiority over the reinforcement learning-based methods in both computation time and energy loss reduction of the hybrid battery system, highlighting the use of such an approach in future energy management systems.

Deep reinforcement learning-based energy management of hybrid battery systems in electric vehicles

The Department of Mechanical Engineering (https:// www.mccormick.northwestern.edu/mechanical/graduate) prepares graduates for careers in industry, research, and academia. Students specialize in Design/Manufacturing/Tribology, Dynamics, Control, Robotics, Neural Engineering, Simulation Driven Engineering, Solid Mechanics, Fluid Dynamics, Nanotechnology/MEMS, Bioengineering and Biomechanics, Energy and Sustainability, Engineering Design and Innovation, Product Development, Engineering Management and other disciplines.

Margaret Lucas

Papers

What is Mechanical Engineering