Real-Time, Three-Dimensional Object Detection and Modeling in Construction

This paper describes a research effort directed to produce methods to model three-dimensional scenes of construction field objects in real-time that adds valuable data to construction information management systems, as well as equipment navigation systems. For efficiency reasons, typical construction objects are modeled by bounding surfaces using a high-frame rate range sensor, called Flash LADAR. The sensor provides a dense cloud of range points which are segmented and grouped into objects. Algorithms are being developed to accurately detect these objects and model characteristics such as volume, speed, and direction. Initial experiments show the feasibility of this method. The advantages and limitations, and potential solutions to limitations are summarized in this paper.