Control and design of an intelligent agricultural robot
Abstract
This thesis deals with the control and design of an intelligent agricultural robot. Robotic melon harvesting has been undertaken as a case study. An intelligent control structure for a robot performing in the uncertain, unknown and ill-structured agricultural domain was modeled as distributed, autonomous computing modules which communicated through a global accessible blackboard structure. The proposed control architecture was tested and verified by simulating all stages of the data flow for the robotic melon harvester: from sensory input, through transformation to information, planning of tasks, and execution of tasks and modification for dynamic conditions. A CAD workstation was used to plan, model, simulate and evaluate the robot's motions based on simulated real-time sensory input. Performance of the robotic system (i.e., cycle time and percentage of successful cycles) was evaluated by developing simulation models that determined the effect of the many closely related robot design parameters (e.g., number of arms; operational mode; i.e. serial, parallel; actuator speeds and accelerations) and horticultural practices (fruit distribution). Initially design alternatives were evaluated using animated, visual simulation which provided insight into the complex interaction between the different system components. Dynamic, numerical simulation tools were then developed to quantify the many closely related design parameters for varying cultural practices. An algorithm, based on the traveling salesman problem, was developed to determine the most time efficient robot for a two dimensional task. Simulations revealed the sensitivity of the robot's design parameters (operational mode, number of arms, actuator speeds, frame size, and picking time) and horticultural practices (planting distance) on the performance of the robotic melon harvester.
Degree
Ph.D.
Advisors
Miles, Purdue University.
Subject Area
Agricultural engineering|Artificial intelligence
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.