Abstract

India is a country which is made up of multiple traditions and culture which get reflected in every aspect of life. Brick masonry is the traditional material used for construction of walls which are load bearing and also to enclose in RC moment resisting frames. The use of brick masonry is inevitable due to the ease in construction, availability of material and availability of good workmanship. Along with this, referring to the seismicity of the country, low to high magnitude earthquakes are also inevitable. In this regard, the knowledge of seismic behaviour of brick masonry infill walls is a necessary aspect in order to contribute to the growing technology. This knowledge is must if the structures are present in higher seismic zones. In context with the past earthquake damages to buildings, the strength based design currently presented in design codes is being felt to be insufficient for withstanding the dynamic forces induced hence the focus of research is being shifted to performance based design. The knowledge of seismic vulnerability of the RC frame due to the presence of brick masonry infills is necessary for performance based design. In this respect, with proper literature review on the existing research on modeling and analyses of brick masonry infill walls a set of case studies have been carried out. Reliable numerical model for this study is verified and finalised. Pushover analysis is used as a tool for observing the different parameters of capacity of structure. The AEM analysis is compared with conventional FEM strut modeling to understand the applicability of the method to this study. The variation of global capacity of the structure with the change in number of bays and storeys is observed by proper analysis. The effect of openings on the behaviour of the RC frame is observed. The short column and weak storey effects are dealt by considering a case study. The weak storey is assumed to be located at different floors and four analyses are run. The comparison between structure with and without short columns is studied. The differences between structure with and without soft storey are observed. To verify the provisions of current standard code of practice IS 1893 – 2002 a case study is performed and results are obtained. For all these analysis final conclusions and results interpretations are drawn.

Abstract

In the recent past, the impact of remote sensing on various earth science applications has significantly increased due to the launch of a vast number of satellites. Remote Sensing now provides data at various scale and granularity which has led to its wide applications. Due to physical limitations in sensor design, there is an inherent trade off between spatial and spectral resolution of these images. Hence, most sensors aim at requirements of few specific applications. This has lead to the under use of vast amounts of data available for various research tasks. Multi-sensor data fusion aims at integrating sources with different strengths such that the final or fused product has strengths of all individual sources. Multi-sensor data fusion has allowed more accurate use of remote sensing data and also allowed new applications to exploit remote sensing datasets. Data can be fused at various levels - pixel, feature or decision level depending on the task at hand. Specifically, pixel level image fusion has advantages in that it involves the raw input data itself and returns a more useful fused dataset with better specifications than that of input data. This makes pixel level fusion independent of the domain or research task and hence makes it efficient and easy to implement. Numerous algorithms have been developed for pixel level fusion. Most algorithms only aim at improving overall visual quality and hence lead to distortion of spectral properties of individual bands which leaves the fused image unsuitable for image processing tasks such as classification. Here, we present two algorithms that aim at Image Enhancement while preserving spectral information of the participating bands. First, we present GLSI (GaussianLinear Stretching Image) Enhancement technique which aims at improving spatial resolution of multispectral bands using high resolution panchromatic band. GLSI transfers spatial details into each band independent of each other and thus preserves the spectral distribution of each band. GLSI is a window based technique where the local statistics of the window is used to create linear relationships between the corresponding windows in the two different spatial resolution bands. Second, we present PA, a hyperspectral image enhancement technique that uses multispectral data as high spatial resolution input. This method first simulates multispectral bands using hyperspectral bands. Then spatial detail is transferred at multispectral level by doing enhancement of simulated multispectral bands to generate fused or enhanced multispectral bands. Then, fused hyperspectral bands are generated by doing vector decomposition. Results and comparison with baseline algorithms demonstrate that proposed algorithms have the potential of doing image enhancement while preserving spectral characteristics.

Abstract

Robot manipulation in clutter with objects in physical contact remains a challenging problem till date. The challenge is posed by interaction involved among the objects at various levels of complexity. Understanding positional semantics of the environment plays an important role in such tasks. The interaction with surrounding objects in the environment must be considered in order to perform the task without causing the objects fall or get damaged. In our work, we learn the semantics in terms of support relationship among different objects in a cluttered environment by utilizing various photometric and geometric properties of the scene. To manipulate an object of interest, we use the inferred support relationship to derive a sequence in which its surrounding objects should be removed while causing minimal damage to the environment. We believe, this work can push the boundary of robotic applications in grasping, object manipulation and picking-from-bin, towards objects of generic shape and size and scenarios with physical contact and overlap. In the first part of the thesis, we aim at learning semantic interaction among objects of generic shapes and sizes lying in clutter involving both direct and indirect physical contact. Three types of support relationships are inferred: "Support from below", "Support from side", and "Containment". Subsequently, the learned semantic interaction or support relationship is used to derive a sequence or order in which the objects surrounding the object of interest should be removed without causing damage to the environment. The generated sequence is called Support Order. We have proposed and analysed two alternative approaches for support inference. In the first approach "Multiple Object Support Inference", support relations between all possible pairs are inferred. In the second approach "Hierarchical Support Inference", given an object of interest, its support relationship with other graspable objects is inferred hierarchically. The support relationship is used to predict the "support order" or the order in which the surrounding objects need to be removed in order to manipulate the target object. In the second part of the thesis, we attempt to learn the semantic interaction among different objects in clutter using multiple views. At first, support relationship among objects in each view is inferred. Then the inferred support relationships are combined to define support relationships across multiple views. The combined global support relationship is used to recover missing support relations and predict the support order. Support order is the order in which objects surrounding an object of interest should be removed. The support order predicted using global support relationship incorporates hidden objects and missing spatial support relations. We have created two RGBD datasets consisting of various objects used in day-to-day life present in clutter. In "Indoor dataset for clutter", 50 cluttered scenes are captured from frontal view using 35 objects of different shapes and sizes. In "Indoor multiview dataset", 7 cluttered scene are captured. Each scene each captured from multiple views. In this dataset, total 67 images are captured using 9 objects of different shapes and sizes. The dataset is made publicly available for the research community around the world. We explore many different settings involving different kind of object-object interaction. We successfully learn support relationships and predict support order in these settings. It can play significant role in extending the scope of manipulation to cluttered environment involving both direct and indirect physical contact, and generic objects. Keywords: Robotic Vision, Support Relations, Support Order, RGBD, Semantic Interaction, Clutter, Multiple Views

Abstract

In this work we explored various methods in improving the accuracy of dependency parsing for Indian languages. Constraint based parsers have been known to have the power to capture non-trivial language generalizations and are well suited for handling the complex phenomenons found in Morphologically rich free word order languages. On the downside, they suffer from robustness, ambiguity resolution and efficiency issues. Data driven parsers, on the other hand, are efficient, deterministic and quite robust. However, they are generally not good at capturing the complex linguistic generalizations necessary to parse MoR-FWO languages. In this work, we explore both these methods and investigate how insights from Computational Paninian grammar (CPG) can inform both constraint based and datadriven parsing in improving their accuracies and also how these parsing paradigms can benefit from one another. We have successfully implemented the linguistically rich graph based dependency parsing approach. Through a series of experiments we formulated the most optimal constraint graph that gives us the best accuracy. These experiments show that some of the previous MSTParser errors can also be corrected consistently.The baseline UAS was 88.66 and the best result obtained from a series of experiments is an UAS of 89.31. This is an increase of 0.65%. There was also an increase of 0.45% in the LAS. Further we studied the effect of minimal semantics on parsing and applied it on a larger scale using Hindi Wordnet to extract ontological categories and provided them as features to data driven parsers. After successfully implementing the hybrid approach for dependency parsing of Hindi we proceeded to implement the same for Sanskrit. We started by adapting the CBHP to Sanskrit which due to lack of proper resources gave less accuracies. But the data driven parser(MST) gave considerably good results after pruning the partially annotated data. The best result achieved is 70.90% in UAS and 82.47% in LAS. We were able to achieve the state-of-the-art accuracy for dependency parsing of Sanskrit by using the partially annotated data.

Abstract

Control of quantum phenomenon using lasers has been a long standing problem since their inception in 1960's. There have been several approaches that have been developed to control processes at a molecular level. Among them, optimal control theory has been one of the most successful schemes developed to achieve the desired objectives. There were other method like the well known Krotov method, the Zhu-Rabitz method and a recently formulated two-point boundary-value quantum control paradigm has gone into developing efficient schemes to deal with the iterative process of searching for optimized fields. Our focus in this thesis has been on using OCT for electronic control. A recent exper- iment by Iwakura, Yabushita and Kobayashi have demonstrated an alternative approach i.e initiation of a pericyclic reaction in the electronic state by means of an ultrashort laser pulse in the UV range. In accordance with this work, Manz et al [1] have used optimal control theory to design a laser pulse to initiate a pericyclic reaction in the ground state. In cognate with this approach, we use an excited state to mediate a reaction on a ground state electronic potential surface and use optimal control theory in combination with time- dependent Schrodinger equation to generate electric field to achieve the desired product. The optimization of the field was carried out using the gradient descent technique. We have managed to achieve an overlap of about 20%. Quantum control is intimately connected to quantum information. Researchers have been trying to use optimal control for performing quantum information protocols e.g. the use of optimal control theory to build quantum gates has been explored by [2]. We explore some aspects of quantum information, in particular "state-merging". In this protocol, mul- tiple parties share a mixed quantum state. The objective here is to merge the state to one of the parties. In the case of two parties A and B, the cost of state merging is given by S(A|B). When we generalise the situation to three parties A,B and C, the cost of state- merging is given by S(A|B)+S(C|A,B). The pertinent question we ask is whether the cost of state merging can be reduced by performing measurement on the system. The answer is negative in the case of two parties as shown by [3] i.e measurement can only increase the cost of state merging. In fact they showed that this change in the cost of state-merging is exactly equal to discord which is always positive. But the situation in case of three parties is completely different. We make a deeper analysis of this situation in the thesis. We find that the change in the cost of state merging is equal to D2 - D1 where D1 is dissension and D2 is discord. Our results clarify that the change in cost of state merging is dependent upon the competition between dissension and discord. In addition we study the phenomenon of dissension in more detail. We define two tracks for the definition of dissension. The first track is the usual way of defining dissension as the difference between the quantum and classical versions of mutual information as mentioned in [4].We have introduced a second track of defining dissension from the perspective of considering all possible measurements. These approaches not only encompass measures of correlation like discord but also give a unified view of the quantum correlation from the view point of projective measurement done on the subsystem.

Abstract

Schistosomiasis is a tropical disease caused by a trematode parasite Schistosoma mansoni that affects more than 240 million people every year and results in large number of deaths annually. It is reported that Thioredoxin Glutathione Reductase (TGR) is a key flavoenzyme present in the schistosomes that bridges the two major redox systems (Thioredoxin and Glutathione), crucial for cellular redox homeostasis and survival of parasite inside the host. The unusual unique metabolic arrangement of thiol-based detoxification pathway found in the parasitic flatworms, has been proposed as a promising drug target. To aid in the development of an efficient drug that target SmTGR, it is essential to understand the structural details, function and redox mechanism of this multifunctional enzyme. Different experiments have been reported, corresponding to various crystal structures, under different biochemical conditions. On the basis of the available structures, their biochemical conditions of crystallization and the general formulation of the events involved in the functioning of related oxidoreductases, Thioredoxin Reductase (TR) and Glutathione Reductase (GR), we have proposed the structural driving force for the different steps involved in the redox mechanism of the enzyme. We have studied different events occurring during the charge transfer process and linked them with the possible redox states of the enzymes on the basis of available crystal structures. We made a good progress in terms of elucidating the structure of the complete dimer of SmTGR and later a cogent hypothesis linking possible structures corresponding to redox states was made. To validate our hypothesis, we have adopted different computational formalisms to investigate the interplay of different redox states of the cofactors with the conformational and physicochemical variations of the neighboring active site residues facilitating the charge transfer process in SmTGR. Firstly, we had used continuum electrostatics and Monte Carlo method to analyze the protonation behaviour of FAD and other titratable residues surrounding the active site.However, the results were ambiguous at physiological pH due to limitations of the methodology used for the calculations. We also modeled the charge transfer pathway from NADPH to FAD using QC/MM studies. Results obtained from QC/MM studies, including spatial distribution of HOMO-LUMO orbitals, ESP charges, etc.; enable us (a) to characterize the nature of the charge transfer from NADPH to FAD and (b) to formulate the possible charge transfer pathway from NADPH to FAD. We further propose that, charge transfer from NADPH to FAD possibly becomes easier with assisted proton coupled movement from NADPH to Glu300 and from Lys162 to FAD.

Abstract

In this thesis, the development of a novel compliant modular robot is proposed for deploy- ment in Urban Search And Rescue(USAR) scenarios. In the light of the recent natural disasters in India and Japan, there is a renewed interest in developing rough terrain robots that can aid in search and rescue efforts by navigating through the rubble to search for trapped victims and dispense basic first-aid. Specifically, USAR scenarios are very challenging for robots. Each disaster is unique and the terrain knowledge gained from one may not be useful in the other. Additionally, the obstacles are so varied in size, shape and rigidity that a single block mobile robot is often limited in its ability to traverse due to either its size or its limited internal freedom to naturally deform along the obstacles. Keeping this in view, we propose the use of a modular design. Each robot module consists of a link and a wheel pair, that are capable of traversing on their own. The wheels are powered to provide propulsion for the robot. The modules are combined to form a snake-like structure and a joint, called link joint, is created between every module pair to allow free rotation (pitch) in the sagittal plane. In the literature, these type of robots are called snake-like robots and more specifically Active Wheel-Passive Joint (AW-PJ) type of robots. These robots typically have a passive articulation mechanism due to the absence of motors at the link joints to control the internal configuration of the robot. While these robots have better ability to navigate through tight pockets, they are also limited in their ability to climb heights greater than one module length due to the problem of tip-over. The primary contribution of this thesis is to enable these robots to climb bigger obstacles and develop a methodology to estimate the number of modules required to climb any given height. Initial work focused on the development of a semi-active modular robot. The robot had five modules and it was minimally equipped with two motors, one each at the second and fourth link joints, respectively. The idea was to use the passive mechanism for climbing steps of smaller heights and motor only while climbing steps of bigger heights. A model-based controller was designed for the active link joints to achieve big-step climbing. The robot therefore had the ad- vantage of height-independent climbing motion as in the case of passive mechanism along with the extra freedom of active joints for maintaining vehicle stability, when necessary. Efficacy of the mechanism was exhibited through simulations on steps of various heights. However, it was realized that this approach was useful to climb heights only upto twice the individual module length. To climb bigger heights, the robot would need motors at all its link joints, leading to a more bulkier design. Generally, motors at the link joints are also susceptible to damage due to impact loads during traversal, making such designs less durable. In order to overcome this disadvantage, the use of compliant joints is proposed. Motors are no longer used at the link joints for posture control. Instead, springs are designed such that they are stiff enough to restrict modules from tipping over while offering sufficient compliance to maneuver on uneven terrains. Spring stiffness for such joints is estimated by formulating an optimization problem over the static equilibrium equations of the robot. The optimization ob- jective is to minimize all link joint moments at every instant of its climbing maneuver. This is one of the key novelties of the proposed work. A design methodology is also proposed for de- veloping an n-module compliant robot for climbing given height on a surface with prescribed coefficient of friction. The performance of the proposed formulation is illustrated for climbing big obstacles and traversing uneven terrains, using both numerical and experimental validation on 3- and 5-module robots. The robot is successfully able to climb maximum heights of 17 cm and 36 cm using 3 and 5 modules, respectively. As a secondary contribution, an optimal wheel-torque based controller is developed to min- imize wheel slip. This helps in reducing odometric error and maximizing energy efficiency. Wheel actuators are the only actively controllable elements of this robot, as the modules are connected using passive compliant joints. Wheel-slip offers a lot of hindrance while traversing on uneven terrains. Therefore, minimizing wheel-slip is crucial to augment its performance. Here, the optimization objective is to minimize the traction-to-normal force ratios at all the wheels. It is shown both numerically and experimentally that the proposed controller not only minimizes slip but also reduces the mean torque requirement for traversing on uneven terrains.

Abstract

Tweets are at most 140 character messages sent by millions of users and dissemination services to provide almost real-time state of affairs in the world starting from an individual level to the issues affecting the complete world. Tweets can be processed to determine events, trends, and to provide multiple services. In this thesis, we address the problem of determining the concepts covered in a tweet automatically, and do away with any manual inspection for various tweet mining tasks. Annotation of tweets with concepts can also be useful for different analyses on tweets at the semantic level such as event analysis, trend analysis etc. We present SCAT, a system which does not require any training data, to annotate tweets with concepts by utilizing the available context information along with the named entities present in it. We present the results of our solution by associating relevant Wikipedia articles and words from WordNet to a tweet. SCAT is able to expand the 140 character tweet with concepts from Wikipedia and WordNet for further semantic processing to state what the tweet is all about. We show that the concepts with which the tweets are annotated using SCAT is able to match the true concepts of the tweets most of the time.

Abstract

In places of reverence, wherein large crowds gather to have small duration of time for individual solace, there is a long queue of people waiting for their turn. A typical example of this scenario is a temple wherein each person visiting the temple, would want to spend a small amount of time to offer prayers in front of a deity. During auspicious occasions there will be hundreds of thousands or more anxious people queuing at the temple wanting to pray. Some other examples would be long shopping queues and queuing at ticketing counters for large crowd gathering events (like sports, music concerts). A crowd management force (that is police or security) at such events, are present to monitor the crowd and prevent any crowd disasters. Yet, there have been cases of stampedes with significant loss of life and trauma in such occasions because of lack of efficiency in the management of crowds. This presents us with a scenario where a multi-agent based system can be used to help improve the efficiency of crowd management at temples. In this thesis, we develop a multi-agent based system for simulating the scenario of crowding in long queues at temples. We focus on how introducing external agents in this scenario would help manage the crowd better. First, we introduce MAMAC (Multi-Agent MAnagement of Crowd), a set of robotic agents that can move at a height above the queue, (i) to provide direction and monitor the crowds, and (iii) to avoid occurrences of stampedes. We then focus on increasing the efficiency of the robotic agents. A method of movement has been developed that can be used by these robotic agents to ensure smooth flow of the crowd. Finally, we present a solution that uses only the required number of robotic agents that can manage the crowd to achieve a stampede free flow. We evaluate the efficiency of crowd management by conducting various experiments on the multi-agent simulation showing the results of robotic-agents avoiding the possibility of stampedes to occur.

Abstract

Online websites have become an important source of information. Most of the information pertaining to different fields of interest can be searched on the World Wide Web. Given the many number of websites relevant for a topic of interest, which one will be preferred by the user, is the question. Research has shown that the usability and the perception of credibility are key factors that make a website successful and influence the user's preference for choosing and using a website. Usability refers to how easy a website is to use. A website with a high degree of usability allows faster and easier navigation, while Credibility (or the perception of Credibility) refers to the degree to which a website could be believed upon. In this research we have focused on these two factors. The research problems addressed are i) How can the Credibility and Usability of website be assessed? ii) How could we automate the process of usability and credibility assessment, so as to benefit the user and/or designer? We started with a review of available usability testing methods and focused on the Cognitive Models (CoLiDeS - Comprehension-based Linked Model of Deliberate Search and its variants). These models use latent semantic analysis technique for finding the similarity between user goal and the website content. The Cognitive Walk-through for Web (CWW) method is a usability inspection method that allows objective evaluation of a website by predicting navigation patterns and finding problematic headings and links. We applied the CWW using CoLiDeS+Pic (Comprehension-based Linked Model of Deliberate Search with Pictures) and found the usability problems (ambiguous links) for a given website. The results can be validated using behavioral studies. The usability problems identified using the CWW simulation can be compared with the actual user navigation patterns observed. We illustrate an example pilot study for the same. Further we focused on the moment of processing of pictures on web pages using eye tracking taking the position of pictures and relevance to the website content into account. The motivation behind this study was that the cognitive model CoLiDeS+Pic used previously for usability inspection, assumes that pictures are processed during the Attention phase (which is the first processing phase). The results showed the assumption of CoLiDeS+Pic model holds true. It was also found that the relevance and the position of pictures on the webpage play an important role in how they are processed. Left side and highly relevant pictures are paid more attention by the user. Thus, the results of this experiment are helpful in making a website better, by having an adequate usage of pictures on the same. As more usable websites allow faster navigation, an automated navigation support tool was developed which is based on a combination of two cognitive models CoLiDeS+Pic (considers pictures) and CoLiDeS+ (considers structural dimensions). The tool is partially automated as the semantics of pictures have still to be obtained manually. An experiment was also conducted to measure the effectiveness of the tool. A significant difference was observed in terms of accuracy, (dis)orientation and time needed to perform the search tasks, between tool-supported and no-support groups in favor of the support tool. Apart from the usability of a website, credibility is also a pre-requisite for its success. So we further studied the user's perception and various factors influencing credibility, based on the previous research. An automated tool was developed to generate credibility scores for websites based on few factors: type of website, date of last updation, sentiments, web traffic data, reputation and review data. We evaluated the tool by relating the tool generated scores and ratings of human judges. A significant positive correlation was observed, which indicates that the tool is compatible with the credibility judgments of users. In the concluding remarks we discussed further improvements in this research area, which include complete automation of the navigation support tool, incorporating more credibility factors and user feedback through machine learning techniques to improve the performance of the credibility assessment tool.