Object-Level Footprint Uncertainty Quantification in Infrastructure-Based Sensing

We examine the problem of analytically estimating uncertainty of objects captured using infrastructure-based camera sensing. Specifically, we aim to derive a closed-form relationship between the ground coordinates and the sources of the camera errors. Using the error propagation equation, the covariance of a given ground coordinate can be measured as a function of the camera errors. The uncertainty of the footprint of the bounding box can then be given as the function of all the extreme points of the object footprint. To calculate the uncertainty of a ground point, the typical error sizes of the error sources are required. We present a method of estimating the typical error sizes from an experiment using a static, high-precision LiDAR as the ground truth. Finally, we present a simulated case study and a real data collection of uncertainty quantification (UQ) from infrastructure-based camera—for simulation, we use CARLA to provide a sense of how the uncertainty changes across a left turn maneuver with two cameras placed with orthogonal field-of-view, and for real data, we examine straight, and right turn maneuvers for a GPS instrumented vehicle.