The Planetary Systems Imager (PSI) is a modular instrument optimized for direct imaging and characterization of exoplanet and disks with the Thirty Meter Telescope (TMT). PSI will operate across a wide wavelength range (≈0.6 - 5μm) to image exoplanets and circumstellar disks in both reflected light and thermal emission. Thanks to the TMT's large collecting area, PSI will have the sensitivity to directly image and spectrally characterize large gaseous planets with unprecedented sensitivity. PSI will also be capable of imaging rocky planets in the habitable zones of the nearest M-type stars in reflected light and search for biomarkers in their atmospheres. Imaging habitable planets in reflected light is PSI's most challenging goal, requiring high contrast imaging (HCI) capabilities well beyond what current instruments achieve. This science goal drives PSI's wavefront sensing and control requirements and defines the corresponding architecture discussed in this paper. We show that PSI must deliver 1e-5 image contrast ≈15 mas separation at λ ≈ 1μm-1.5μm, and that a conventional extreme-AO architecture relying on a single high speed wavefront sensor (WFS) is not sufficient to meet this requirement. We propose a wavefront control architecture relying on both visible light (λ < 1.1 μm) sensing to optimize sensitivity, and near-IR (λ > 1.1 μm) sensors to address wavefront chromaticity terms and provide high contrast imaging capability. We show that this combination will enable speckle halo suppression at the < 1e-5 raw contrast level in near-IR, allowing detection and spectroscopic characterization of potentially habitable exoplanets orbiting nearby M-type stars.