The development of a geo-referencing system for machine controlled construction equipment

Machine controlled construction equipment (MCE) continues to garner support from the construction industry due to shortages of skilled labor, constant technological advances and the importance of construction to the overall economy. MCE requires reliable, real-time geo-referencing of the equipment end-effector. However, MCE research continues to focus on relative positioning for control system design, while research dealing with geo-referencing has been mainly in the fields of aerial mapping and terrestrial mobile mapping, which can benefit from post-processing. The research described in this thesis attempts to overcome this problem by developing a real-time geo-referencing system specifically for MCE. The total system consists of three components; an integrated DGPS/INS to geo-reference the equipment main-body; an open kinematic chain to relatively position the end-effector with respect to the main-body; and a unified model to geo-reference the end-effector. The system carrier is designed for an excavator, but the model for the development of the open kinematic chain, designed using the Denavit-Hartenberg convention, can accommodate any type of MCE. The overall objective was the development of a precise geo-referencing system that could be operated under all construction conditions and could achieve an accuracy required for the recording of exposed infrastructure which calls for a vertical component of 15mm. This required high-level accuracy in both the position and orientation, therefore, DGPS and INS were integrated. Furthermore, the positional accuracy required double-differenced carrier phase measurements implemented using a least squares method for ambiguity resolution. Extended Kalman filters (EKF) were used for DGPS baseline estimation and DGPS/INS integration, the latter using a tightly-coupled, closed-loop architecture. Finally, error analysis was completed on the open kinematic chain to resolve the accuracy required in the joint sensors. System testing was completed using sample data from an International Federation of Surveyors Commission mobile van test and simulated data for the open kinematic chain. Results showed that the geo-referencing system was able to achieve ±0.024m (RMSE) horizontally and ±0.034m (RMSE) in height when the excavator was stationary and executing a common digging trajectory. The accuracy achieved would allow the excavator to operate autonomously for several common construction tasks.