FROM SUB-MEGAPIXEL TO MULTI-MEGAPIXEL RESOLUTION: A SCALABLE DMD-PLM HYBRID ToF SOLID-STATE LIDAR WITH DIFFRACTIVE AND HOLOGRAPHIC BEAM STEERING WITH CROSSTALK MITIGATION

Abstract

This thesis discusses a novel hybrid optical LiDAR architecture that employs Texas Instruments’ Digital Micromirror Devices (DMDs) for coarse beam and image steering, a Phase Light Modulator (PLM) for fine field of view steering in a Time-of-Flight (ToF) solid-state LiDAR with contiguous panoramic LiDAR scanning and higher resolution imaging.By synchronizing the laser pulse to the dynamic tilt movement of the micromirrors on the DMD, a blazed grating condition can be satisfied, and light is diffracted and steered into one of the several diffraction orders with high diffraction efficiency. By employing Computer Generated Holograms (CGHs) on the PLM, the micro-mirrors’ height on the PLM can be modulated based on the hologram’s parameters, and a variety of grating patterns can be created for controlled image and/or field of view steering. There are multiple hierarchies of FoVs in this paper. For simplicity, the system’s total field of view is defined as the FOV of the system, the subdivisions of FOVs from the diffractive beam steering by DMDs are termed as ‘Sub-FOV’, and the finer FOVs within the sub-FOV of the DMD are termed as ‘Sub-Sub-FOV’. The proposed DMD-PLM Hybridized scanning and solid-state flash LiDAR architecture utilizes a near-infrared nanosecond pulsed laser, two synchronized DMDs (one for ‘Transmitter’ and the other for ‘Receiver’) for coarse sub-FOV beam steering, a PLM for finer sub-sub-FOV steering, and a Multi-Pixel Photon Counter (MPPC) to capture two-dimensional ToF LiDAR images. The experimental results demonstrate that the proposed LiDAR architecture increases the effective pixel count by 9-fold in a single sub-FOV by employing PLM sub-sub-FOV steering. There are seven diffraction orders from the DMDs used for coarse steering in the experiment, and each DMD order’s sub-FOV carries 9 sub-sub-FOVs from the PLM fine steering. Therefore, there is a 9-fold increase in the effective pixel count of the original LiDAR architecture, which only employed DMDs for the beam and image steering [2]. An addition of a holographic-based FOV steering PLM increases the pixel resolution by multiple folds in the LiDAR imaging by fine steering into the sub-FOV regions. With advantages, there exist some challenges in this hybrid LiDAR as well. There is the presence of the strong specular reflections at the 0th order due to the cover glass reflection from both the DMDs and PLM, as well as PLM’s 0th order interference with its higher orders from PLM’s MEMS mirror due to non-linearity and truncated phase modulation in the infrared region. The protective cover glass layer of both the DMDs and PLM is VIS/UV coated but not coated for infrared, causing strong Fresnel reflection at the 0th order. This unmodulated reflection introduces crosstalk between the 0th order and higher orders, making it difficult to distinguish higher-order diffraction images from the zeroth-order image. To mitigate this issue, a physical Fourier mask is applied to block the cover glass reflection from the transmitter-DMD, while a polarization-selective Fourier filtering technique is used to suppress the PLM’s cover-glass reflection and its zeroth order. This thesis presents a further advancement in the hybridized scanning and flash LiDAR architecture described by Chan (2023) [2]. The proposed new architecture enables an n²-fold enhancement in pixel resolution over Chan’s architecture of DMD-only based LiDAR by employing phase modulation with a PLM to perform sub-sub-FOV steering within each DMD’s sub-FOV. Along with the pixel resolution enhancement, this new architecture addresses the crosstalk issues caused by the cover glass specular reflection, beam spilling on adjacent orders, and gaps in beam steering and LiDAR imaging, utilizing a paraxial raytrace model for illumination design of the transmitter and contiguous receiver FOV design. Both the theoretical formulation and experimental validation of the resolution enhancement mechanism are provided. The crosstalk suppression techniques are effectively handled using Fourier-domain masking for DMD’s cover-glass reflection and polarization-selective filtering for the PLM’s cover glass and 0th order. The system is further supported by an analytical first-order model for contiguous, gap-free LiDAR imaging, and also demonstrates foveated scanning capabilities. These advancements offer significant potential for high-resolution, compact, and adaptive LiDAR systems in applications such as autonomous driving, robotics, and 3D mapping.