Abstract

This paper introduces geometry and novel object shape and pose costs for multi-object tracking in road scenes. Using images from a monocular camera alone, we devise pairwise costs for object tracks, based on several 3D cues such as object pose, shape, and motion. The proposed costs are agnostic to the data association method and can be incorporated into any optimization framework to output the pairwise data associations. These costs are easy to implement, can be computed in real-time, and complement each other to account for possible errors in a tracking-by-detection framework. We perform an extensive analysis of the designed costs and empirically demonstrate consistent improvement over the state-of-the-art under varying conditions that employ a range of object detectors, exhibit a variety in camera and object motions and, more importantly, are not reliant on the choice of the association framework. We also show that, by using the simplest of associations frameworks (two-frame Hungarian assignment),we surpass the state-of-the-art in multi-object-tracking on road scenes

Abstract

We present a new paradigm for real-time objectoriented SLAM with a monocular camera. Contrary to previous approaches, that rely on object-level models, we construct category-level models from CAD collections which are now widely available. To alleviate the need for huge amounts of labeled data, we develop a rendering pipeline that enables synthesis of large datasets from a limited amount of manually labeled data. Using data thus synthesized, we learn categorylevel models for object deformations in 3D, as well as discriminative object features in 2D. These category models are instance-independent and aid in the design of object landmark observations that can be incorporated into a generic monocular SLAM framework. Where typical object-SLAM approaches usually solve only for object and camera poses, we also estimate object shape on-the-fly, allowing for a wide range of objects from the category to be present in the scene. Moreover, since our 2D object features are learned discriminatively, the proposed object-SLAM system succeeds in several scenarios where sparse feature-based monocular SLAM fails due to insufficient features or parallax. Also, the proposed category models help in object instance retrieval, useful for Augmented Reality (AR) applications. We evaluate the proposed framework on multiple challenging real-world scenes and show — to the best of our knowledge — first results of an instance-independent monocular object-SLAM system and the benefits it enjoys over feature-based SLAM methods.

Abstract

In this paper, we present a novel control law for longitudinal speed control of autonomous vehicles. The key contributions of the proposed work include the design of a control law that reactively integrates the longitudinal surface gradient of road into its operation. In contrast to the existing works, we found that integrating the path gradient into the control framework improves the speed tracking efficacy. Since the control law is implemented over a shrinking domain scheme, it minimizes the integrated error by recomputing the control inputs at every discretized step and on sequently provides less reaction time. This makes our control law suitable for motion planning frameworks that are operating at high frequencies. Furthermore, our work is implemented using a generalized vehicle model and can be easily extended to other classes of vehicles. The performance of gradient aware - shrinking domain based controller is implemented and tested on a stock electric vehicle on which a number of sensors are mounted. Results from the tests show the robustness of our control law for speed tracking on a terrain with varying gradient while also considering stringent time constraints imposed by the planning framework.

Abstract

This paper focuses on the customized-wireless sensor node implementation of the classical least mean square (LMS) algorithm for the reduction in data-transmissions from the sensor nodes to the sink in internet of things (IoT) networks. This reduction, in turn, increases the battery life of the sensor node. The system was deployed in outdoor and indoor environments to read the ambient temperature and then perform the prediction of the sensed data in order to minimize the number of data-transmissions to the sink node. The utility of the proposed concept has been demonstrated using the measured data and the battery life is increased 2.64 and 2.53 times in indoor and outdoor environments, respectively.

Abstract

In buildings, a large chunk of energy is spent on heating, ventilation and air conditioning systems. One way to optimize their usage is to make them demand-driven depending on human occupancy. This paper focuses on accurately estimating the number of occupants in a room by leveraging multiple heterogeneous sensor nodes and machine learning models. For this purpose, low-cost and non-intrusive sensors such as CO 2 , temperature, illumination, sound and motion were used. The sensor nodes were deployed in a room in a star configuration and measurements were recorded for a period of four days. A regression based method is proposed for calculating the slope of CO 2 , a new feature derived from real-time CO 2 values. Supervised learning algorithms such as linear discriminant analysis (LDA), quadratic discriminant analysis (QDA), support vector machine (SVM) and random forest (RF) were used on several different combinations of feature sets. Moreover, multiple performance metrics such as accuracy, F1 score and confusion matrix were used to evaluate the performance of our models. Experimental results demonstrate a maximum accuracy of 98.4% and a high F1 score of 0.953 for estimating the number of occupants in the room. Principal component analysis (PCA) was also applied to evaluate the performance of a dataset with reduced dimensionality.

Abstract

The analyzed system model in this paper is a distributed parallel detection network in which each secondary user (SU) evaluates the energy-based test statistic from the received observations and sends it to a fusion center (FC), which makes the final decision. Uncertainty in the noise variance at each SU is modeled as an unknown constant in a certain interval around the nominal noise variance. It is assumed that the SUs are heterogeneous in that the nominal noise variances and the uncertainty intervals can be different for different SUs. Moreover, the received signal power at each SU may be different. For the considered system model, the paper presents important results for two inter-related themes on cooperative energy detection (CED) in the presence of noise uncertainty (NU). First, the expressions for generalized signal-to-noise ratio (SNR) walls are derived for the classical CED fusion rule, i.e., sum of energies from all SUs. Second, a Dempster-Shafer theory based CED is proposed in the presence of NU with heterogeneous sensors. In the proposed scheme, the test statistic from each SU is the energy-based basic mass assignment values, which are first discounted depending on the uncertainty level associated with the SU and then fused at the FC using the Dempster rule of combination to arrive at the global decision. It is shown that the proposed scheme outperforms the traditional sum fusion rule in terms of detection performance as well as the location of SNR wall.

Abstract

With the success of deep learning based approaches in tackling challenging problems in computer vision, a wide range of deep architectures have recently been proposed for the task of visual odometry (VO) estimation. Most of these proposed solutions rely on supervision, which requires the acquisition of precise ground-truth camera pose information, collected using expensive motion capture systems or high-precision IMU/GPS sensor rigs. In this work, we propose an unsupervised paradigm for deep visual odometry learning. We show that using a noisy teacher, which could be a standard VO pipeline, and by designing a loss term that enforces geometric consistency of the trajectory, we can train accurate deep models for VO that do not require ground-truth labels. We leverage geometry as a selfsupervisory signal and propose "Composite Transformation Constraints (CTCs)", that automatically generate supervisory signals for training and enforce geometric consistency in the VO estimate. We also present a method of characterizing the uncertainty in VO estimates thus obtained. To evaluate our VO pipeline, we present exhaustive ablation studies that demonstrate the efficacy of end-to-end, self-supervised methodologies to train deep models for monocular VO. We show that leveraging concepts from geometry and incorporating them into the training of a recurrent neural network results in performance competitive to supervised deep VO methods.

Abstract

The focus of this paper is on evaluating different spatial interpolation methods for the construction of radio environment map using field measurements obtained by cyclostationary-based mobile sensors. Mobile sensing devices employing cyclostationary detectors provide lot of advantages compared to the widely used energy detectors, such as robustness to noise uncertainty and ability to distinguish among different primary user signals. However, mobile sensing results are not available at locations between the sensors making it difficult for a secondary user (possibly without a spectrum sensor) to decide whether to use primary user resources at that location. To overcome this, spatial interpolation of test statistics measured at limited number of locations can be carried out to create a channel occupancy map at unmeasured locations between the sensors. For this purpose, different spatial interpolation techniques for the cyclostationary test statistic have been employed in this paper such as inverse distance weighting, ordinary kriging, and universal kriging. The effectiveness of these methods is demonstrated by applying them on extensive real-world field measurement data obtained by mobile-phone-compliant spectrum sensors. The field measurements were carried out using four mobile spectrum sensors measuring eight digital video broadcasting-terrestrial (DVB-T) channels at more than hundred locations encompassing roughly one-third of the area of the city of Espoo in Finland. The accuracy of the spatial interpolation results based on the field measurements is determined using the cross-validation approach with the widely used root mean square error as the metric. Field measurement results indicate that reliable results with spatial coverage can be achieved using kriging for cyclostationary based test statistics. Comparison of spatial interpolation results of cyclostationary test statistics is also carried out with those of energy values obtained during the measurement campaign in the form of received signal strength indicator. Comparison results clearly show the performance improvement and robustness obtained using cyclostationary based detectors instead of energy detectors.

Abstract

A master equation has been constructed for a global system–bath interaction in the both absence and presence of non-Markovian noise. For the memoryless case, it has been exactly solved for a paradigmatic class of two qubit states in high- and zero-temperature thermal environment. For the non-Markovian model, it has been solved for zero-temperature bath. The evolution of quantum coherence and entanglement has been observed in the presence of the above-mentioned interactions. We show that the global part of the system–bath interaction compensates for the decoherence, resulting in slowdown of coherence and entanglement decay. For an appropriately defined limiting case, both coherence and entanglement show freezing behavior for the high-temperature bath. In case of zero-temperature bath, the mentioned interaction not only stabilizes the non-classical correlations, but also enhances them …

Abstract

The design and fabrication of a rolled dielectric elastomer actuator is described and the parametric dependence of the displacement and blocked force on the actuator geometry, elastomer layer thickness, voltage, and number of turns is analyzed. Combinations of different elastomers and carbon nanotube electrodes are investigated and optimized to meet performance characteristics appropriate to tactile display applications, namely operation up to 200 Hz with a combination of a 1 N blocked force and free displacement of 1 mm, all within a volume of less than 1 cm3. Lives in excess of 50 000 cycles have been obtained. Key to meeting these objectives is control of the multilayering fabrication process, the carbon nanotube electrode concentration, the selection of a soft elastomer with low viscous losses, and a proof‐testing procedure for enhancing life cycle reliability.