Advanced tracking techniques for Active Target Time Projection Chamber detectors in Nuclear Physics experiments

Tracking detectors for low-energy nuclear physics experiments, acting at the same time as reaction targets, are very promising devices in a wide range of research topics. The ability to measure and reconstruct the trajectory of all of the reaction products with high efficiency and good geometrical resolution allows particle spectroscopy studies to be performed under experimental conditions below the sensitivity threshold of standard techniques. This is possible only if solid, efficient, and computationally fast reconstruction codes are implemented and validated. In the presented thesis project, reconstruction codes and classification techniques have been developed aimed at processing experimental data from the ACTAR Active Target. The purpose of such codes was to provide accurate information about track geometry, particle energy, and identification. In particular, the reaction $^{20}$O(d,$^3$He)$^{19}$N$^\star$ has been first analyzed using the Hough transform and RANSAC algorithms, comparing their efficiencies. In the second part, the results obtained applying machine learning techniques on the same data will be presented, with the aim of achieving a fast event classification employing for the first time on ACTAR these cutting-edge techniques.