Autonomous ground-based robotic navigation for an agricultural row crop environment

The research presented in this thesis is focused on the navigation technique for an autonomous ground-based robotic system for use in an agricultural row crop environment. The performance of the navigational system was evaluated by measuring the offset of the robot’s path to a predetermined path. It is found through a total of ten field tests that utilizing highly accurate GPS systems results in the greatest navigation accuracy, with an average offset of 3.56-inches. During the agricultural growing season, many row crops produce a canopy that restricts the ability to observe and measure the various atmospheric and biological processes that take place beneath the canopy, affect the various growth stages of the plant, and ultimately alter the crop yield. The increasing use of unmanned aerial vehicles (UAV) for agricultural purposes has increased our ability to monitor crop growth. Mainly due to cost limitations, however, most studies on the interaction of environmental factors on plant growth are focused on end-point measurements. Ground-based robotic technologies provide a new method for obtaining measurements that give insight into the effect of environmental factors that affect plants during many different stages of a plant’s growth cycle. Furthermore, much more frequent analysis and modeling of the crops can be obtained using a ground-based robotic approach. This allows for more accurate yield estimations as the great number of varying conditions make yield estimations derived from fewer measurements much more difficult and complex. One of the greatest drawbacks to using a UAV approach to monitor and estimate crop growth and yield is the lack of sensing in the sub-canopy environment. Other drawbacks include the necessity for high-cost localization hardware used to facilitate navigation. In order to overcome the limitations of aerial-based sensing, this work proposes a low-cost ground-based solution for sub-canopy monitoring. This research focuses specifically on the rover navigation technique, which is a main aspect in the foundation of the proposed project. iii The navigation method employed in this study was evaluated in both laboratory and agricultural settings. This was, in part, an effort to help simulate the various terrain and environmental conditions that may be experienced in a real life setting. Utilizing various types of navigation methods, the ability of each method to successfully navigate through the rows of a field was quantified by analyzing the deviation from the ideal path, or a straight line, as commonly seen in row crop settings. A total of ten straight-line tests were conducted, each with slightly different navigational parameters and configurations. GPS waypoints were used to instruct the robot to drive in a straight line for 10-meter segments. The results of this study indicate that a Real Time Kinematic (RTK) GPS system provides the greatest accuracy and ability for row crop navigation, with an average offset from the desired path of 3.56-inches. This solution also provides an opportunity for applying ground-based navigational solutions for various projects that may require frequent and detailed measurements obtained by on-board sensors. This research is important to researchers because it provides a low-cost autonomous robotic navigational system that can be used in a wide range of projects, such as the continuous monitoring of the sub-canopy environment of a row crop field