Abstract:The uncertainty disturbance is one of the main disturbances that seriously influences the stabilization precision of an aerial inertially stabilized platform (ISP) system. In this paper, to improve the stabilization precision of the ISP under disturbance uncertainty, a robust H∞ controller is designed in this paper. Then, the reduction order is carried out for high-order controllers generated by the robust H∞ loop shaping control method. The application of the minimum implementation and balanced truncation algorithm in controller reduction is discussed. First, the principle of reduced order of minimum implementation and balanced truncation are analyzed. Then, the method is used to reduce the order of the high-order robust H∞ loop shaping controller. Finally, the method is analyzed and verified by the simulations and experiments. The results show that by the reduced-order method of minimum implementation and balanced truncation, the stabilization precision of the robust H∞ loop shaping controller is increased by about 10%.

Abstract:The Wireless Sensor Networks (WSNs) are widely utilized in various industrial and environmental monitoring applications. The process of data gathering within the WSN is significant in terms of reporting the environmental data. However, it might occur that certain sensor node malfunctions due to the energy draining out or unexpected damage. Therefore, the collected data may become inaccurate or incomplete. Focusing on the spatiotemporal correlation among sensor nodes, this paper proposes a novel algorithm to predict the value of the missing or inaccurate data and predict the future data in replacement of certain nonfunctional sensor nodes. The Long-Short-Term-Memory Recurrent Neural Network (LSTM RNN) helps to more accurately derive the time-series data corresponding to the sets of past collected data, making the prediction results more reliable. It is observed from the simulation results that the proposed algorithm provides an outstanding data gathering efficiency while ensuring the data accuracy.

Abstract:Hydroponic farming is a viable and economical farming method, which can produce safe and healthy greens and vegetables conveniently and at a relatively low cost. It is essential to provide supplemental lighting for crops grown in greenhouses to meet the daily light requirement, Daily Light Integral (DLI). The present paper investigates how effectively and efficiently LEDs can be used as a light source in hydroponics. It is important for a hydroponic grower to assess the requirement of photosynthetically active radiation (PAR) or the Photosynthetic Photon Flux Density (PPFD), in a greenhouse, and adjust the quality and quantity of supplemental lighting accordingly. A Quantum sensor (or PAR sensor) can measure PAR more accurately than a digital light meter, which measures the light intensity or illuminance in the SI unit Lux, but a PAR sensor is relatively expensive and normally not affordable by an ordinary farmer. Therefore, based on the present investigation and experimental results, a very simple way to convert light intensity measured with a Lux meter into PAR is proposed, using a simple conversion factor (41.75 according to the present work). This allows a small-scale hydroponic farmer to use a simple and inexpensive technique to assess the day to day DLI values of PAR in a greenhouse accurately using just an inexpensive light meter. The present paper also proposes a more efficient way of using LED light panels in a hydroponic system. By moving the LED light panels closer to the crop, LED light source can use a fewer number of LEDs to produce the same required daily light requirement and can increase the efficiency of the power usage to more than 80%. Specifically, the present work has determined that it is important to design more efficient vertically movable LED light panels with capabilities of switching individual LEDs on and off, for the use in greenhouses. This allows a user to control the number of LEDs that can be lit at a particular time, as required. By doing so it is possible to increase the efficiency of a LED lighting system by reducing its cost of the electricity usage.

Abstract:This paper concerns the data-driven sensor deployment problem in large spatiotemporal fields. Traditionally, sensor deployment strategies have been heavily dependent on model-based planning approaches. However, model-based approaches do not typically maximize the information gain in the field, which tend to generate less effective sampling locations and lead to high reconstruction error. In the present paper, a data-driven approach is developed to overcome the drawbacks of the model-based approach and improve the spatiotemporal field reconstruction accuracy. The proposed method can select the most informative sampling locations to represent the entire spatiotemporal field. To this end, the proposed method decomposes the spatiotemporal field using principal component analysis (PCA) and finds the top r essential entities of the principal basis. The corresponding sampling locations of the selected entities are regarded as the sensor deployment locations. The observations collected at the selected sensor deployment locations can then be used to reconstruct the spatiotemporal field, accurately. Results are demonstrated using a National Oceanic and Atmospheric Administration sea surface temperature dataset. In the present study, the proposed method achieved the lowest reconstruction error among all methods.

Abstract:In this paper, negative selection and genetic algorithms are combined and an improved bi-objective optimization scheme is presented to achieve optimized negative selection algorithm detectors. The main aim of the optimal detector generation technique is maximal nonself space coverage with reduced number of diversified detectors. Conventionally, researchers opted clonal selection based optimization methods to achieve the maximal nonself coverage milestone; however, detectors cloning process results in generation of redundant similar detectors and inefficient detector distribution in nonself space. In approach proposed in the present paper, the maximal nonself space coverage is associated with bi-objective optimization criteria including minimization of the detector overlap and maximization of the diversity factor of the detectors. In the proposed methodology, a novel diversity factor-based approach is presented to obtain diversified detector distribution in the nonself space. The concept of diversified detector distribution is studied for detector coverage with 2-dimensional pentagram and spiral self-patterns. Furthermore, the feasibility of the developed fault detection methodology is tested the fault detection of induction motor inner race and outer race bearings.

Abstract:The main content of a news web page is a source of data for Natural Language Processing (NLP) and Information Retrieval (IR), which contains large quantities of valuable information. This paper proposes a method that formulates the main content extraction problem as a DOM tree node classification problem. In terms of feature extraction, we use the DOM tree node to represent HTML document and then develop multiple features by using the DOM tree node properties, such as text length, tag path, tag properties and so on. In consideration that the essence of the problem is the classification model, we use Xgboost to help select nodes. Experimental results show that the proposed approach is effective and efficient in extracting main content of new web pages, and achieves about 98% accuracy over 1083 news pages from 10 different new sites, and the average processing time per page is within 10ms.

Abstract:In this research work, a hierarchical controller has been designed for an autonomous navigation robot to avoid unexpected moving obstacles where the state and action spaces are continuous. The proposed scheme consists of two parts: 1) a controller with a high-level approximate reinforcement learning (ARL) technique for choosing an optimal trajectory in autonomous navigation; and 2) a low-level, appearance-based visual servoing (ABVS) controller which controls and execute the motion of the robot. A novel approach for path planning and visual servoing has been proposed by the combined system framework. The characteristics of the on-board camera which is equipped on the robot is naturally suitable for conducting the reinforcement learning algorithm. Regarding the ARL controller, the computational overhead is quite low thanks to the fact that a knowledge of obstacle motion is not necessary. The developed scheme has been implemented and validated in a simulation system of obstacle avoidance. It is noted that findings of the proposed method are successfully verified by obtaining an optimal robotic plan motion strategy.

Abstract:Images that are taken underwater mostly present color shift with hazy effects due to the special property of water. Underwater image enhancement methods are proposed to handle this issue. However, their enhancement results are only evaluated on a small number of underwater images. The lack of a sufficiently large and diverse dataset for efficient evaluation of underwater image enhancement methods provokes the present paper. The present paper proposes an organized method to synthesize diverse underwater images, which can function as a benchmark dataset. The present synthesis is based on the underwater image formation model, which describes the physical degradation process. The indoor RGB-D image dataset is used as the seed for underwater style image generation. The ambient light is simulated based on the statistical mean value of real-world underwater images. Attenuation coefficients for diverse water types are carefully selected. Finally, in total 14490 underwater images of 10 water types are synthesized. Based on the synthesized database, state-of-the-art image enhancement methods are appropriately evaluated. Besides, the large diverse underwater image database is beneficial in the development of learning-based methods.

Abstract:The use of hand gestures can be the most intuitive human-machine interaction medium. The early approaches for hand gesture recognition used device-based methods. These methods use mechanical or optical sensors attached to a glove or markers, which hinder the natural human-machine communication. On the other hand, vision-based methods are less restrictive and allow for a more spontaneous communication without the need of an intermediary between human and machine. Therefore, vision gesture recognition has been a popular area of research for the past thirty years. Hand gesture recognition finds its application in many areas, particularly the automotive industry where advanced automotive human-machine interface (HMI) designers are using gesture recognition to improve driver and vehicle safety. However, technology advances go beyond active/passive safety and into convenience and comfort. In this context, one of America’s big three automakers has partnered with the Centre of Pattern Analysis and Machine Intelligence (CPAMI) at the University of Waterloo to investigate expanding their product segment through machine learning to provide an increased driver convenience and comfort with the particular application of hand gesture recognition for autonomous car parking. The present paper leverages the state-of-the-art deep learning and optimization techniques to develop a vision-based multiview dynamic hand gesture recognizer for a self-parking system. We propose a 3D-CNN gesture model architecture that we train on a publicly available hand gesture database. We apply transfer learning methods to fine-tune the pre-trained gesture model on custom-made data, which significantly improves the proposed system performance in a real world environment. We adapt the architecture of end-to-end solution to expand the state-of-the-art video classifier from a single image as input (fed by monocular camera) to a Multiview 360 feed, offered by a six cameras module. Finally, we optimize the proposed solution to work on a limited resource embedded platform (Nvidia Jetson TX2) that is used by automakers for vehicle-based features, without sacrificing the accuracy robustness and real time functionality of the system.

Abstract:The intersection of Quantum Technologies and Robotics Autonomy is explored in the present paper. The two areas are brought together in establishing an interdisciplinary interface that contributes to advancing the field of system autonomy, and pushes the engineering boundaries beyond the existing techniques. The present research adopts the experimental aspects of quantum entanglement and quantum cryptography, and integrates these established quantum capabilities into distributed robotic platforms, to explore the possibility of achieving increased autonomy for the control of multi-agent robotic systems engaged in cooperative tasks. Experimental quantum capabilities are realized by producing single photons (using spontaneous parametric down-conversion process), polarization of photons, detecting vertical and horizontal polarizations, and single photon detecting/counting. Specifically, such quantum aspects are implemented on network of classical agents, i.e., classical aerial and ground robots/unmanned systems. With respect to classical systems for robotic applications, leveraging quantum technology is expected to lead to guaranteed security, very fast control and communication, and unparalleled quantum capabilities such as entanglement and quantum superposition that will enable novel applications.