Structure learning of graphical models for task-oriented robot grasping

In the collective imaginaries a robot is a human like machine as any androids in science fiction. However the type of robots that you will encounter most frequently are machinery that do work that is too dangerous, boring or onerous. Most of the robots in the world are of this type. They can be found in auto, medical, manufacturing and space industries. Therefore a robot is a system that contains sensors, control systems, manipulators, power supplies and software all working together to perform a task. The development and use of such a system is an active area of research and one of the main problems is the development of interaction skills with the surrounding environment, which include the ability to grasp objects. To perform this task the robot needs to sense the environment and acquire the object informations, physical attributes that may influence a grasp. Humans can solve this grasping problem easily due to their past experiences, that is why many researchers are approaching it from a machine learning perspective finding grasp of an object using information of already known objects. But humans can select the best grasp amongst a vast repertoire not only considering the physical attributes of the object to grasp but even to obtain a certain effect. This is why in our case the study in the area of robot manipulation is focused on grasping and integrating symbolic tasks with data gained through sensors. The learning model is based on Bayesian Network to encode the statistical dependencies between the data collected by the sensors and the symbolic task. This data representation has several advantages. It allows to take into account the uncertainty of the real world, allowing to deal with sensor noise, encodes notion of causality and provides an unified network for learning. Since the network is actually implemented and based on the human expert knowledge, it is very interesting to implement an automated method to learn the structure as in the future more tasks and object features can be introduced and a complex network design based only on human expert knowledge can become unreliable. Since structure learning algorithms presents some weaknesses, the goal of this thesis is to analyze real data used in the network modeled by the human expert, implement a feasible structure learning approach and compare the results with the network designed by the expert in order to possibly enhance it.

Tecniche avanzate per l'analisi di sensitività globale: applicazione alla contaminazione delle acque sotterranee.

The thesis is framed within the field of the stochastic approach to flow and transport themes of solutes in natural porous materials. The methodology used to characterise the uncertainty associated with the modular predictions is completely general and can be reproduced in various contexts. The theme of the research includes the following among its main objectives: (a) the development of a Global Sensitivity Analysis on contaminant transport models in the subsoil to research the effects of the uncertainty of the most important parameters; (b) the application of advanced techniques, such as Polynomial Chaos Expansion (PCE), for obtaining surrogate models starting from those which conduct traditionally developed analyses in the context of Monte Carlo simulations, characterised by an often not negligible computational burden; (c) the analyses and the understanding of the key processes at the basis of the transport of solutes in natural porous materials using the aforementioned technical and analysis resources. In the complete picture, the thesis looks at the application of a Continuous Injection transport model of contaminants, of the PCE technique which has already been developed and applied by the thesis supervisors, by way of numerical code, to a Slug Injection model. The methodology was applied to the aforementioned model with original contribution deriving from surrogate models with various degrees of approximation and developing a Global Sensitivity Analysis aimed at the determination of Sobol’ indices.