The search of exoplanets using gravitational microlensing started in 1995. First results come quickly and the first exoplanet has been detected on 2003 by the OGLE and the MOA collaborations. Nowadays, more than twenty exoplanets are published and the number of detections should rise a lot in the coming years with the launch of KMTNet survey and EUCLID and WFIRST space-based observatories. When a "close" star, the lens, cross the line of sight between the Earth and a more distant star, the source, the luminosity of this source is magnified. By chance, if a planet orbits this lens star, it produces a second magnification of lower amplitude. The lightcurve of the event then shows a typical signature : a planetary deviation. In this manuscript, we present the theoritical and observational tools needed to detect planets with gravitational microlensing. We present two studies on two special events : MOA-2009-BLG-411L, a lens formed by a brown dwarf around an M-dwarf, and MOA-2010-BLG-477Lb, a super-Jupiter orbiting an M star. We finally present a study on planets already published in order to standardize theses results. Two major problems make this research difficult. First, the occurence of a microlensing event for a single star is very low (one in a million). We then need to observe very dense fields, such as the Galactic Bulge. Each night, OGLE and MOA surveys analyze the Galactic Bulge to detect events. The second problem is the planetary deviation is very short, one hour to few days for most massive planets. This is why we have to continously observe events and use telescopes all around the Southern Hemisphere. The rapid increase of detection during the last years forces the follow-up telescopes to choose which targets they need to observe. We so decided to develop a new software to make automatically this choice. At the end of this manuscript, we present a statistical study of microlensing parameters from four observational seasons. We expect these results will help to constrain theoritical model of the Milky Way in the future.
