Energy is essential for human existence and our future depends on plentiful and accessible sources of energy. The world population is fast growing and the average energy used per capita increases. One of the greatest challenges for human beings is that of meeting the growing demand for energy in a responsible, equitable and sustainable way. The possibility to obtain energy by "fusing" light atoms addresses these needs. Nuclear fusion reactions are clean, safe and the amount of fuel present on Earth (hydrogen isotopes) is practically inexhaustible and well distributed. Nuclear fusion is a natural process that occurs in all active stars like our Sun. Since the first demonstration of a deuterium fusion reaction (Rutherford 1933), researchers worldwide have tried to replicate this process on Earth by building a thermonuclear fusion reactor. Nevertheless, the challenge posed by the construction of a nuclear fusion reactor is greater than the one presented earlier by the development of a fission reactor. During the IAEA Conference in Geneva in the early 1958, L.A.Artsimovich declared: "Plasma physics is very difficult. Worldwide collaboration is needed for progress" and E.Teller, at the same conference: "Fusion technology is very complex. It is almost impossible to build a fusion reactor in this century". They were right. The extremely high temperature and density necessary to fuse hydrogen isotopes makes it difficult indeed to create a successful fusion reactor. Even though the physics of the fusion reaction appears clear, we are still facing problems on the road towards bulding the "box" that can efficiently confine the hot gas in the state of plasma. The best results so far have been obtained confining a plasma with strong magnetic fields in a toroidal configuration ("tokamak"). The Centre de Recherches en Physique des Plasmas in Switzerland actively studies this promising configuration towards the development of a nuclear fusion reactor. The experimental activity of the Tokamak à Configuration Variable (TCV) mainly focuses on the research of optimized plasma shapes capable of improving the global performance and solve the technological challenges of a tokamak reactor. Several theoretical and experimental results show the importance of the plasma shape in tokamaks. The maximum value of β (an indicator of the confinement efficiency) is for example related to the ratio between the height and the width of the plasma. The plasma shape can also affect the power necessary to access improved confinement regimes, as well as the plasma stability. This thesis reports on a contribution towards the optimization of the tokamak plasma shape. In particular, it describes the theoretical and experimental studies carried out in the TCV tokamak on two innovative plasma shapes: the doublet shaped plasma and the snowflake divertor. Doublet shaped plasmas have been studied in the past by the General Atomics group. Since then, the development of new plasma diagnostics and the discovery of new confinement regimes have given new reasons for interest in this unusual configuration. TCV is the only tokamak worldwide theoretically able to establish and control this configuration. This thesis illustrates new motivations for creating doublet plasmas. The vertical stability of the configuration is studied using a rigid model and the results are compared with those obtained with the KINX MHD stability code. The best strategy for controlling a doublet on TCV is also investigated, and a possible setup of the TCV control system is suggested for the doublet configuration. Analyzing the possible scenarios for doublet creation, the most promising scenario consists of the creation of two independent plasmas, which are subsequently merged to establish a doublet. For this reason, particular attention needs to be devoted to the problem of the plasma start-up. In this thesis, a general analysis of the TCV ohmic and assisted with ECH plasma start-up is presented, and recent attempts to create a doublet plasma are reported. Since the magnetic field reconstruction at the breakdown time is important to better diagnose these plasmas, the entire magnetic system of TCV has been calibrated with an original technique, also described in the manuscript. The last part of this thesis is devoted to the snowflake divertor configuration. This innovative plasma shape has been proposed and theoretically studied by Dr. D.D.Ryutov from the Lawrence Livermore National Laboratory. In Ryutov's articles, this configuration was proposed to alleviate the problems of the plasma-wall interaction and possibly affect the plasma edge stability. The TCV tokamak was the first to report the creation and control of a snowflake configuration, and the candidate was the principal investigator of this work. These results are accordingly discussed in this thesis. Details are provided in particular on the strategy used to establish the configuration. An edge-localized mode (ELM) H-mode regime, supported by electron cyclotron heating, has been successfully established in a snowflake. This regime exhibits 2 to 3 times lower ELM frequency but only a 20%-30% increase in normalized ELM energy (ΔWELM/WP ) compared to an identically-shaped, conventional, single-null, diverted H-mode. Enhanced stability of mid- to high-toroidal-mode-number ideal modes is consistent with the different snowflake ELM phenomenology. Finally, the capability of the snowflake to redistribute the edge power on the additional strike points has been confirmed experimentally and is also reported in this thesis.