I have worked on direct detection and detailed characterization of exoplanets using extreme adaptive optics (ExAO) as a key science case of both current and future telescopes. However, both quasi-static and residual atmospheric wavefront errors currently limit the sensitivity of this endeavour, generating \speckles" in a coronagraphic image that initially obscure any faint exoplanet(s) from detection. I demonstrated the current limits of exoplanet imaging using datasets taken with the Gemini Planet Imager and Subaru Coronagraphic ExAO systems. Even when using advanced post-processing algorithms, speckle evolution over time and wavelength was shown to limit the final contrasts that can be reached with current state-of-the-art instruments. A new approach was thus needed to detect fainter exoplanets below these limits.I then illustrated a path forward to reach near-photon noise-limited contrasts: fast focal plane wavefront sensing of both quasi-static and atmospheric speckles. My new method, called the Fast Atmospheric Self-coherent camera Technique (FAST), was designed precisely to overcome these limitations. Looking toward the future, the contrast improvements from fast focal plane wavefront sensing techniques such as FAST are expected to play an essential role in the ground-based detection and characterization of lower mass exoplanets.