The advantage of large optical telescopes from the ground for high-contrast direct imaging relies on achieving exquisite control of the incident optical wavefront. Since the exoplanet problem almost always has a bright star within a small field of view there are usually ample photons for measuring the atmospheric and telescopedistorted wavefront. A 5-year ERA Chair-funded Laboratory for Innovation in Opto-Mechanics (LIOM) has just begun at the Instituto de Astrof´ısica de Canarias (IAC). Measuring the wavefront using Machine Learning techniques in combination with a photonic lantern offers the possibility of fully common-mode wavefront sensing for combining the Fizeau telescope sub-apertures to minimize scattered light and to perform accurate wavefront nulling ("dark spot" or "vortex" nulling) in the image plane. Here we describe our progress in implementing this approach through numerical and laboratory simulation.