Tommaso Campari

Ritratto Tommaso Campari

Computer Science and Innovation for Societal Challenges, XXXV series
Grant sponsor

Fondazione Bruno Kessler
Luciano Serafini (FBK), Lamberto Ballan


Project description
Virtually every action we take every day, from the simplest to the most complex, can be reduced to a sequence of sub-actions that lead from a starting state to a goal. Unconsciously when we plan how to carry out an action we rely on our sight and our previous experiences. If the task to be solved is totally new to us, we will try to solve it and in the future, we will try to learn from our mistakes. My research project is born of these intrinsic characteristics in human behavior and my purpose is to develop a vision-based planning agent that mimics the way humans plan. The agent will be based on an abstract representation of the environment in which it operates, which will be continuously updated through the use of incremental learning techniques. For this purpose it will use sensors suitable for capturing images and through technologies for their analysis taken from the field of Computer Vision, it will be able to extract information from them. This will allow to have an abstract representation of the environment in continuous evolution and, possibly, to start from an empty representation that will be expanded step by step. This will be possible thanks to all the new technologies presented in recent years in the field of artificial intelligence. In fact, today various architectures are available for image and video analysis. These are mainly based on convolutional neural networks and allow to extract a series of crucial information able to describe an image or a frame.