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.

Distribution of metals in several vineyards of North Italy: from soil to wine

Nowadays we live in densely populated regions and this leads to many environmental issues. Among all pollutants that human activities originate, metals are relevant because they can be potentially toxic for most of living beings. We studied the fate of Cd, Cr, Cu, Fe, Mn, Ni, Pb and Zn in a vineyard environment analysing samples of plant, wine and soil. Sites were chosen considering the type of wine produced, the type of cultivation (both organic and conventional agriculture) and the geographic location. We took vineyards that cultivate the same grape variety, the Trebbiano). We investigated 5 vineyards located in the Ravenna district (Italy): two on the Lamone Valley slopes, one in the area of river-bank deposits near Ravenna city, then a farm near Lugo and one near Bagnacavallo in interfluve regions. We carried out a very detailed characterization of soils in the sites, including the analysis of: pH, electric conductivity, texture, total carbonate and extimated content of dolomite, active carbonate, iron from ammonium oxalate, Iron Deficiency Chlorosis Index (IDCI), total nitrogen and organic carbon, available phosphorous, available potassium and Cation Exchange Capacity (CEC). Then we made the analysis of the bulk chemical composition and a DTPA extraction to determine the available fraction of elements in soils. All the sites have proper ground to cultivate, with already a good amount of nutrients, such as not needing strong fertilisations, but a vineyard on hills suffers from iron deficiency chlorosis due to the high level of active carbonate. We found some soils with much silica and little calcium oxide that confirm the marly sandstone substratum, while other soils have more calcium oxide and more aluminium oxide that confirm the argillaceous marlstone substratum. We found some critical situations, such as high concentrations of Chromium, especially in the farm near Lugo, and we noticed differences between organic vineyards and conventional ones: the conventional ones have a higher enrichment in soils of some metals (Copper and Zinc). Each metal accumulates differently in every single part of grapevines. We found differences between hill plants and lowland ones: behaviors of plants in metal accumulations seems to have patterns. Metals are more abundant in barks, then in leaves or sometimes in roots. Plants seem trying to remove excesses of metal storing them in bark. Two wines have excess of acetic acid and one conventional farm produces wine with content of Zinc over the Italian law limit. We already found evidence of high values relating them with uncontaminated environments, but more investigations are suggested to link those values to their anthropogenic supplies.