| This thesis presents a framework for an intelligent earthwork system (IES). The proposed framework defines the architecture and methodologies to serve as the foundation for developing an IES. The IES enables multiple pieces of equipment to automatically perform field operations such as earthmoving and landfill operations. The main goals of this research are (1) to develop a construction agent model for IES, (2) to develop a task identification and planning method for effective task execution, (3) to develop a resource allocation method in order to maximize equipment utilization, and (4) to develop a dynamic path planning algorithm to avoid collisions in the construction site.; IES consists of a task planning sub-system (TPS), a task execution sub-system (TES), and a human control sub-system (HCS), which each have two or more agents. All agents in the system have autonomy and interact with each other in a partially centralized and a partially decentralized way. Hybrid control approach is suggested for the IES and the function layers of the IES control approach are presented. The flow of information between functions is modeled.; The task identification and planning in IES involve the translation of input data to the first-level tasks that may consist of more than one subtask, and the explanation of what will happen with the scheduled time zones, where it will happen, and how it will happen. To generate the schedule of field operation tasks, a heuristic rule, Task Package Sequencing Rule (TPSR), is developed.; The resource allocation policies are developed, which is based on the contract net algorithm. The concept of an Agent Cluster (AC) is presented and six factors used for resource allocation are identified. Based on identified factors, an equation model for bidding value calculation is suggested.; This thesis describes SensBug algorithm, which generates a path for point mobile equipment in a 2D plane unknown environment populated by both stationary and movable obstacles. The SensBug algorithm can overcome the shortcoming of the previous dynamic path planning algorithm. |
