Optical Beamforming Techniques for Solid-State Automotive LIDAR

Pavan Bhargava

EECS Department
University of California, Berkeley
Technical Report No. UCB/EECS-2023-30
May 1, 2023

http://www2.eecs.berkeley.edu/Pubs/TechRpts/2023/EECS-2023-30.pdf

The advent of new products in advanced driver assistance systems, virtual reality headsets, robotics, and biometrics has led to increasing demand for high resolution 3D depth maps. LIDAR (Light Detection and Ranging) has emerged as a strong candidate sensor to supply these applications with depth maps of the environment. In fact, LIDAR is already used as a primary sensor in experimental self-driving vehicles, since it provides high-resolution 3D images that existing RADAR solutions do not. However, the reliability and cost of these automotive LIDAR sensors has precluded their use in high-volume mainstream applications.

This thesis explores and demonstrates beamsteering architectures that significantly reduce the size, weight, power and cost of LIDAR sensors. In particular, we focus on designs which achieve the high-volume, reliability, and resolution that automotive applications demand. We begin with an overview of existing LIDAR sensor architectures, and show that the key to improving cost and scalability is switching from traditional mechanical beam-steering to solid-state beam-steering using silicon photonic platforms. We then model and analyze the performance of different solid-state beam steering choices in the context of automotive applications, and arrive at a design for a thermal optical phased array. This thesis then outlines silicon implementation results from from the first 3D integrated the LIDAR SoC based on the aforementioned thermal optical phased array design. Next, we analyze several key issues that prevent this prototype from scaling to more demanding specifications, and propose a new beamsteering architecture based on focal-plane arrays which circumvents these fundamental issues. Finally, the models and automated electronic-photonic design methodologies developed in this work culminate in the design of a scalable monolithically integrated focal-plane array transmitter in a zero-change 45nm SOI CMOS process.

Advisor: Vladimir Stojanovic


BibTeX citation:

@phdthesis{Bhargava:EECS-2023-30,
    Author = {Bhargava, Pavan},
    Editor = {Stojanovic, Vladimir},
    Title = {Optical Beamforming Techniques for Solid-State Automotive LIDAR},
    School = {EECS Department, University of California, Berkeley},
    Year = {2023},
    Month = {May},
    URL = {http://www2.eecs.berkeley.edu/Pubs/TechRpts/2023/EECS-2023-30.html},
    Number = {UCB/EECS-2023-30},
    Abstract = {The advent of new products in advanced driver assistance systems, virtual reality headsets, robotics, and biometrics has led to increasing demand for high resolution 3D depth maps. LIDAR (Light Detection and Ranging) has emerged as a strong candidate sensor to supply these applications with depth maps of the environment. In fact, LIDAR is already used as a primary sensor in experimental self-driving vehicles, since it provides high-resolution 3D images that existing RADAR solutions do not. However, the reliability and cost of these automotive LIDAR sensors has precluded their use in high-volume mainstream applications.

This thesis explores and demonstrates beamsteering architectures that significantly reduce the size, weight, power and cost of LIDAR sensors. In particular, we focus on designs which achieve the high-volume, reliability, and resolution that automotive applications demand. We begin with an overview of existing LIDAR sensor architectures, and show that the key to improving cost and scalability is switching from traditional mechanical beam-steering to solid-state beam-steering using silicon photonic platforms. We then model and analyze the performance of different solid-state beam steering choices in the context of automotive applications, and arrive at a design for a thermal optical phased array. This thesis then outlines silicon implementation results from from the first 3D integrated the LIDAR SoC based on the aforementioned thermal optical phased array design. Next, we analyze several key issues that prevent this prototype from scaling to more demanding specifications, and propose a new beamsteering architecture based on focal-plane arrays which circumvents these fundamental issues. Finally, the models and automated electronic-photonic design methodologies developed in this work culminate in the design of a scalable monolithically integrated focal-plane array transmitter in a zero-change 45nm SOI CMOS process.}
}

EndNote citation:

%0 Thesis
%A Bhargava, Pavan
%E Stojanovic, Vladimir
%T Optical Beamforming Techniques for Solid-State Automotive LIDAR
%I EECS Department, University of California, Berkeley
%D 2023
%8 May 1
%@ UCB/EECS-2023-30
%U http://www2.eecs.berkeley.edu/Pubs/TechRpts/2023/EECS-2023-30.html
%F Bhargava:EECS-2023-30