In this paper we develop a model capable of classifying drivers from their driving behaviors sensed by only low level sensors. The sensing platform consists data available the diagnostic outlet (OBD) car and smartphone We window based support vector machine to classify drivers. test our with two datasets collected under both controlled naturalistic conditions. Furthermore, evaluate using each sensor source (car phone) independently combining average classification accuracies attained three...

In recent years, mobile devices have become a part of our daily lives--much like television sets had over the second half 20th century. Increasingly, people are using while watching television. We set out to understand this behavior on minute-by-minute quantified level as well users' motivations and purposes device use conducted novel mixed-methods study inside seven households with fourteen instrumented phone tablet devices, capturing all app launches durations, correlated moment in program...

PbS/PVP nanocomposite films with tunable band gap from 0.8 to 1.92 eV.

Activity recognition is a core component of many intelligent and context-aware systems. In this paper, we present solution for discreetly unobtrusively recognizing common work activities above surface without using cameras. We demonstrate our approach, which utilizes an RF-radar sensor mounted under the surface, in two domains; at convenience-store counter (useful post-hoc analytics) office deskwork real-time applications). classify seven clerk with 94.9% accuracy data collected lab...

This letter proposes a new image-based localization framework that explicitly localizes the camera/robot by fusing Convolutional Neural Network (CNN) and sequential images' geometric constraints. The camera is localized using single or few observed images training with 6-degree-of-freedom pose labels. A Siamese network structure adopted to train an image descriptor network, visually similar candidate in set retrieved localize testing geometrically. Meanwhile, probabilistic motion model...

Convolutional Neural Networks (CNN) have successfully been utilized for localization using a single monocular image [1]. Most of the work to date has either focused on reducing dimensionality data better learning parameters during training or developing different variations CNN models improve pose estimation. Many best performing works solely consider content in image, while context from historical images is ignored. In this paper, we propose combined CNN-LSTM which capable incorporating...

In this paper, we propose a novel solution to optimize the deployment of Radio Frequency (RF) beacons for purpose indoor localization. We system that optimizes both number and their placement in given environment. cost-function, called CovBsm, allows simultaneously 3-coverage while maximizing beacon spreading. Using cost function, framework maximize The proposed accounts infrastructure its influence on RF signal propagation by embedding realistic simulator into optimization process.

Sliding-window-based low-rank matrix approximation (LRMA) is a technique widely used in hyperspectral images (HSIs) denoising or completion. However, the uncertainty quantification of restored HSI has not been addressed to date. Accurate denoised facilitates applications such as multisource multiscale data fusion, assimilation, and product since these require an accurate approach describe statistical distributions input data. Therefore, we propose prior-free closed-form element-wise method...

The electrochemical supercapacitor has shown to be a reliable and innovative type of energy storage technology over the years. Recent research that CoWO4 is potential material for applications because its unique characteristics, which make it suitable storage. particles are synthesized using hydrothermal method with PVP as capping agent, followed by 2-hour post-heat treatment at 300 ℃. powder was characterized through X-ray diffraction (XRD), scanning electron microscopy, Raman spectroscopy,...

This paper proposes a Hierarchical Hidden Markov Model (HHMM) framework as the most suitable tool to exploit interactions between an intelligent mobility aid and their human operator. The presented is capable of learning mixed array Activities Daily Living (ADL) that typical user these supportive devices would normally engage in, both navigational non-navigational in nature, provide assistance when required. main contribution this demonstration how probabilistic modelling behaviours at...

A motivating, core capability of most smart, Internet Things enabled spaces (e.g., home, office, hospital, factory) is the ability to leverage context use. Location a key element; particularly indoor location. Recent advances in radio ranging technologies, such as 802.11-2016 FTM, promise availability low-cost, near-ubiquitous time-of-flight-based estimates. In this paper, we build on prior work enhance technology's provide useful location We demonstrate meaningful improvements...

Fatigue Detection Using Computer Vision Long duration driving is a significant cause of fatigue related accidents cars, airplanes, trains and other means transport. This paper presents design detection system which can be used to detect in drivers. The based on computer vision with main focus eye blink rate. We propose an algorithm for that conducted through process extracting the face image from video followed by evaluating region then eventually detecting iris using binary image. advantage...

This paper presents the design of an intelligent walking aid for frail and elderly as well patients who are recovering from surgical procedures, in order to enhance safer mobility these study populations. The device augments a conventional rolling walker with sensing navigational abilities safely travel through environment following user's perceived intentions, unless collisions or instability is imminent. agent, embodied partially observable Markov decision process (POMDP), critically...

In this paper we use a Hierarchical Hidden Markov Model (HHMM) to represent and learn complex activities/task performed by humans/robots in everyday life. Action primitives are used as grammar human behaviour the interactions of human/robots with different objects. The main contribution is probabilistic model capable representing behaviours at multiple levels abstraction support proposed hypothesis. hierarchical nature allows decomposition task into simple action primitives. framework...

The synthesis and characterization are described for the complex salt [Ru(κ2-2,2′-N2C10H6{(CH2)3(CF2)5CF3}2-4,4′)(κ2-2,2′-N2C10H8)2][PF6]2, where one of 2,2′-bipyridyl ligands bears terminal bis(perfluoralkyl) chains, purpose which is to exploit 'like-dissolves-like' intermolecular forces as an anchoring mechanism in Nafion™ membranes. TBAB-modified-Nafion™ thin films (sulfonic acid protons exchanged NBun4+ cations) were cast or spin-coated onto electrodes glass microscope slips, exposed...