Bounding overwatch

About: Bounding overwatch is a research topic. Over the lifetime, 966 publications have been published within this topic receiving 15156 citations.

A novel autonomous staircase cleaning system with robust 3D-Deep Learning-based perception technique for Area-Coverage

Abstract: Cleaning the staircases is the next big leap for every commercial cleaning robot in order to accomplish a full-fledged cleaning of a constructed buildings. Such an effort could be witnessed in the academic literature where a robotic system can autonomously clean the staircase by ascending. However, none of the existing staircase traversing platforms demonstrated the ability to perform both ascending and descending motion while cleaning, which can significantly improvise the overall robot’s performance. In this paper, we propose a novel reconfigurable cleaning robotic platform called sTetro_plotter, which can perform both ascend and descend motion in the staircases. Pointedly, in this work, we presented a perception framework for the developed robot to traverse on the staircase and perform area coverage autonomously. The framework was constructed with a pointNet++ based feature extractor and classification and regression network to generate a bounding box on the targeted feature. Also, we discussed the process of a staircase descending through tracking the generated bounding box. We implemented a sweeping-based lidar device that can generate a 3D point cloud by sensing its environment. We evaluated the performance of the proposed robot and its perception system through conducting experiments in real-world scenarios. The experimental trials successfully demonstrate the ability of the sTetro_plotter robot to perform autonomous area coverage while traversing on the staircase using the developed perception framework.

Bounding Techniques and Their Application to Simplified Plastic Analysis of Structures

Abstract: In the framework of the simplified analysis methods for elastoplastic analysis problems, the bounding techniques possess an important role. A class of these techniques, based on the so-called perturbation method, are here presented with reference to finite element discretized structures. A general bounding principle is presented and its applications are illustrated by means of numerical examples.

An Efficient Image Categorization Method With Insufficient Training Samples.

Abstract: Image classification is an important part of pattern recognition. With the development of convolutional neural networks (CNNs), many CNN methods are proposed, which have a large number of samples for training, which can have high performance. However, there may exist limited samples in some real-world applications. In order to improve the performance of CNN learning with insufficient samples, this article proposes a new method called the classifier method based on a variational autoencoder (CFVAE), which is comprised of two parts: 1) a standard CNN as a prior classifier and 2) a CNN based on variational autoencoder (VAE) as a posterior classifier. First, the prior classifier is utilized to generate the prior label and information about distributions of latent variables; and the posterior classifier is trained to augment some latent variables from regularized distributions to improve the performance. Second, we also present the uniform objective function of CFVAE and put forward an optimization method based on the stochastic gradient variational Bayes method to solve the objective model. Third, we analyze the feasibility of CFVAE based on Hoeffding's inequality and Chernoff's bounding method. This analysis indicates that the latent variables augmentation method based on regularized latent variables distributions can generate samples fitting well with the distribution of data such that the proposed method can improve the performance of CNN with insufficient samples. Finally, the experiments manifest that our proposed CFVAE can provide more accurate performance than state-of-the-art methods.