Exploiting Uncertainty in Triangulation Light Curtains for Object Tracking and Depth Estimation

Abstract

Active sensing through the use of Adaptive Depth Sensors is a nascent field, with potential in areas such as Advanced driver-assistance systems (ADAS). One such class of sensor is the Triangulation Light Curtain, which was developed in the Illumination and Imaging (ILIM) Lab at CMU. This sensor (comprising of a rolling shutter NIR camera and a galvomirror with a laser) uses a unique imaging strategy that relies on the user providing the depth to be sampled, with the sensor returning the return intensity at said location. Prior work demonstrated effective strategies for local depth estimation, but failed to take into account the physical limitations of the galvomirror, work over long ranges, or exploit the triangulation uncertainty in the sensor. Our goal is this thesis is to demonstrate the effectiveness of this sensor in the ADAS space. We do this by developing planning, control and sensor fusion algorithms that consider the device constraints, and exploit the device’s physical effects. We present those results in this thesis.