In this Thesis I present and discuss the work completed during my three and a half years as a PhD student in the Exoplanet Research Group of the Cavendish Laboratory, University of Cambridge, UK. Most of my work has been in collaboration and partnership with the HARPS3 instrument and the Terra Hunting Experiment.

My focus is on the development of new techniques and technologies that are aimed at aiding the discovery of an ‘Earth-twin’ exoplanet. In the context of this work, I use the term Earth-twin to mean an Earth-massed rocky planet orbiting a Sun-like star at a period of around 300 days. I created a pipeline prototype for fitting planetary models to radial velocity (RV) data. The data can contain any number of random or systematic noise sources, and can be poorly sampled. The analysis is conducted in a nested-sampling Bayesian framework and thus allows for the direct statistical comparison of different planetary models given some data set, and produces full posterior estimation for all the parameters of all the models. I used this analysis technique to test the feasibility of using intense ground-based RV surveys to detect Earth-twins, and to compare the results with typical survey cadences. I found that an intense survey reliably and regularly finds a variety of planets, including the Earth-twins, and out-performs the typical survey cadence.

The major new technology I have developed is an experiment to measure the geometric positions of the pixels of an optical CCD, and the data analysis pipeline to compute the results. In exoplanet science, precise measurements of the Doppler shift of the stellar spectral lines enable us to confirm the presence of planets. However, at some level of precision, our uncertainty of the detector itself starts to inhibit our detection capability. Hence, if we are to be successful in the discovery of low mass planets, we require knowledge of the sub-pixel structure of our detector. I used the analysis scripts to help plan and design an optical experiment which was then built to analyse a large format optical detector and measure the positions of the pixels. I found the simulation of the experiment can measure the pixel positions to a precision of less than 0.001 pixels, but the experiment was plagued with thermal variations and ultimately was not capable of such precise measurements.