Human-assisted rapid workspace modeling for construction equipment operations

This dissertation is concerned with developing new useful methods for rapid local area sensing and 3D modeling and simulation for better planning and control of construction equipment operations in unstructured workspaces. Construction requires real-time information about the workspace, which complicates graphical and image modeling and updating. Rapid 3D positioning and modeling1 can be very useful in construction in order to optimize equipment operations and to significantly improve safety and a remote operator's spatial perception of the workspace. Such models can be used to plan, visualize, optimize and communicate operations before execution. By employing human logic, the method developed in this dissertation, data acquisition time and complicated image processing algorithms are reduced. Breaking down the modeling process into human-assisted elements and automated computer-based elements achieved potentially real-time performance. The quantitative results of the error modeling2 for the laser/camera mounted large scale manipulator validated the overall system designs. This dissertation also examines the potential broader usefulness and applicability of the developed human-assisted rapid graphical workspace modeling method by conducting modeling experiments with several outdoor objects. It is concluded that combining human-assisted graphical workspace modeling with pre-stored CAD models and simple sensors (such as single-axis laser rangefinders and remote video cameras) can significantly reduce modeling time while potentially increasing modeling accuracy. It can also be anticipated that the developed graphical modeling method and the error modeling method can be extended to a broad class of construction automation design, simulation, and graphical control problems, such as generating as-builts, assessing infrastructure conditions, and controlling construction operations. 1 This modeling can be defined in this application as a process to produce 3D graphical representation of a real-world situation or objects in terms of volume, position and orientation. 2 This modeling can be defined as the use of mathematics and statistics to evaluate and manipulate error attributes from the collected data.