Search

In order to do a comprehensive research on multi-agent systems (MASs), it is imperative to consider the dynamic models of the agents, communication relationships between the agents, and control protocols to ensure consensus (Zhao et al., 2022). These are the three main aspects of a collaborative multi-agent system, and they are all crucial to the efficient operation of the system. In this research, we analyze a multi-agent system of unmanned ground and aerial agricultural vehicles operating collaboratively to complete a field task autonomously. Specifically, this research focuses on developing a collaboration model among ground and aerial agricultural robots that ensures consensus of the agents during an autonomous planting operation, and a coordination architecture governing the tasks allocation among the agents, and the exchange of states and information between the agents. Additionally, a 3-dimensional visual robotics simulation was developed to evaluate the developed collaboration models for visualizing how the collaboration of the heterogeneous agents would perform in a real-world setting. Advisor: Santosh K. Pitla

A large increase in productivity could be realized by small increases in efficiency and utilization on grasslands due to their large area. Pre-harvest biomass estimations can help forage producers make better informed management decisions. Producers of dry hay and hay silage can better identify plant growth stages and the optimal time to harvest. Grazers can better determine the most efficient number of animals and stocking density. The goal of this study was to create a novel pre-harvest biomass estimation method utilizing compressed height combined with an unmanned ground vehicle (UGV). A compression plate named a “compression ski” was constructed and mounted on an UGV to take continuous compressed height measurements. Compression ski measurements were made before and after harvest to analyze change in compressed height (ΔCH) as an alternative to pre-harvest compressed heights (CH1) alone. A rising plate meter (RPM) was used to make discrete compressed height measurements and calibrated for the same trial plots. The methods were tested on trial plots of tall fescue (Schedonorus phoenix) and alfalfa (Medicago sativa) which were measured and harvested at approximately 10, 20, and 30-day intervals. Compressed height indicators were correlated with wet yield (kg/ha) and estimation models were made. Methods of estimating dry matter fraction (DMF) were created independently. Wet yield estimation models were combined with DMF estimations to create dry matter yield (DM yield) prediction models. The best compression ski DM yield prediction models had standard errors of 291 kg DM/ha for tall fescue and 491 kg DM/ha for alfalfa.

Unmanned ground vehicles (UGVs) are increasingly being used for civilian and military applications. Passive sensing, such as visible cameras, are being used for navigation and object detection. An additional object of interest in many environments is text. Text information can supplement the autonomy of unmanned ground vehicles. Text most often appears in the environment in the form of road signs and storefront signs. Road hazard information, unmapped route detours and traffic information are available to human drivers through road signs. Premade road maps lack these traffic details, but with text localization the vehicle could fill the information gaps. Leading text localization algorithms achieve ~60% accuracy; however, practical applications are cited to require at least 80% accuracy [49]. The goal of this thesis is to test existing text localization algorithms against challenging scenes, identify the best candidate and optimize it for scenes a UGV would encounter. Promising text localization methods were tested against a custom dataset created to best represent scenes a UGV would encounter. The dataset includes road signs and storefront signs against complex background. The methods tested were adaptive thresholding, the stroke filter and the stroke width transform. A temporal tracking proof of concept was also tested. It tracked text through a series of frames in order to reduce false positives. Best results were obtained using the stroke width transform with temporal tracking which achieved an accuracy of 79%. That level of performance approaches requirements for use in practical applications. Without temporal tracking the stroke width transform yielded an accuracy of 46%. The runtime was 8.9 seconds per image, which is 44.5 times slower than necessary for real-time object tracking. Converting the MATLAB code to C++ and running the text localization on a GPU could provide the necessary speedup.

The problem of roadway navigation and obstacle avoidance for unmanned ground vehicles has typically needed very expensive sensing to operate properly. To reduce the cost of sensing, it is proposed that an algorithm be developed that uses a single visual camera to image the roadway, determine where the lane of travel is in the image, and segment that lane. The algorithm would need to be as accurate as current lane finding algorithms as well as faster than a standard k- means segmentation across the entire image. This algorithm, named IRIS, was developed and tested on several sets of roadway images. The algorithm was tested for its accuracy and speed, and was found to be better than 86% accurate across all data sets for an optimal choice of algorithm parameters. IRIS was also found to be faster than a k-means segmentation across the entire image. IRIS was found to be adequate for fulfilling the design goals for the algorithm. IRIS is a feasible system for lane identification and segmentation, but it is not currently a viable system. More work to increase the speed of the algorithm and the accuracy of lane detection and to extend the inherent lane model to more complex road types is needed. IRIS represents a significant step forward in the single camera roadway perception field.

Current trends in agricultural equipment have led to an increasing degree of autonomy. As the state of the art progresses towards fully autonomous vehicles, it is important to consider assumptions implicit in the design of these vehicles. Current automation in harvesters have led to increased sensing and automation on current combines, but no published research examines the effect of machine size on the viability of the autonomous system. The question this thesis examines is: if a human is no longer required to operate an individual harvester, is it possible to build smaller equipment that is still economically viable? This thesis examines the appropriateness of automating these machines by developing a conceptual model for smaller, fully autonomous harvesters. This model includes the basic mechanical subsystems, a conceptual software design, and an economic model of the total cost of ownership. The result of this conceptual design and analysis is a greater understanding of the role of autonomy in harvest. By comparing machine size, machine function, and the costs to own and operate this equipment, design guidelines for future autonomous systems are better understood. It is possible to build an autonomous harvesting system that can compete with current technologies in both harvest speed and overall cost of ownership.

The objective of this research is to explore and develop new methodologies and techniques to improve UGV mission reliability. This dissertation focuses on two research issues that are critical in the following UGV deployment phases: (1) prior to field deployment to remove design deficiencies; and (2) during field usage to prevent mission failures. Four specific research topics are accomplished. The first topic focuses on simulation-based acceptance testing. A general framework is proposed to integrate dynamic and static simulations. Statistical hypothesis testing is used to compare static and dynamic simulations to determine when a simple static simulation can be used to replace the complex dynamic simulation. Results show that the static simulation can be used when a failure mechanism is not significantly affected by the dynamic characteristics of the vehicle. The remaining research topics aim at prevention of operational failures due to unexpected energy depletion. A model-based Bayesian prediction framework integrated with a dynamic vehicle model is proposed in the second research topic, which improves traditional approaches for estimation and prediction. The Bayesian framework combines mission prior knowledge with real-time measurements for adaptive prediction of end-of-mission energy requirement. Experimental studies were conducted, which validated and demonstrated the advantages of the framework on roads with different surface types and grades. The third research topic, entitled real-time energy reliable path planning, builds upon the above mentioned prediction framework to identify the most energy reliable path in a stochastic network with unknown and correlated arc lengths. Since traditional sequential optimization techniques cannot be directly applied to this problem, a heuristic approach based on two stage exploration/exploitation is proposed to identify the most reliable path. The framework, which minimizes the cost of exploration, outperforms traditional path planning approaches. In the final research topic, the impact of operator driving style on mission energy requirements is investigated using statistical response surface. While the previous topics help with overall mission planning regardless of the operator’s driving style, here, improving the driving style to increase energy availability is studied. The optimal drive cycle that minimizes energy consumption and procedures for reduction of energy consumption are proposed.

Unmanned ground vehicles are increasingly employing a combination of active sensors such as LIDAR with passive sensors like cameras to perform at all levels of perception, which includes detection, recognition and classification. Typical cameras measure the intensity of light at a variety of different wavelengths to classify objects in different areas of an image. A polarimetric camera not only measures intensity of light, but can also determine its state of polarization. The polarization of light is the angle the electric field of the wave of light takes as it travels. A polarimetric camera can identify the state of polarization of the light, which can be used to segment highly polarizing areas in a natural environment, such the surface of water. The polarimetric camera designed and built for this thesis was created with low cost in mind, as commercial polarimetric cameras are very expensive. It uses multiple beam splitters to split incoming light into four machine vision cameras. In front of each machine vision camera is a linear polarizing filter that is set to a specific orientation. Using the data from each camera, the Stokes vector can be calculated on a pixel by pixel basis to determine what areas of the image are more polarized. Test images of various scenes that included running water, standing water, mud, and vehicles showed promise in using polarization data to highlight and identify areas of interest. This data could be used by a UGV to make more informed decisions in an autonomous navigation mode.

Perception is a key feature in how any creature or autonomous system relates to its environment. While there are many types of perception, this thesis focuses on the improvement of the visual robotics perception systems. By implementing a broadband passive sensing system in conjunction with current perception algorithms, this thesis explores scene reconstruction and world modeling. The process involves two main steps. The first is stereo correspondence using block matching algorithms with filtering to improve the quality of this matching process. The disparity maps are then transformed into 3D point clouds. These point clouds are filtered again before the registration process is done. The registration uses a SAC-IA matching technique to align the point clouds with minimum error.  The registered final cloud is then filtered again to smooth and down sample the large amount of data. This process was implemented through software architecture that utilizes Qt, OpenCV, and Point Cloud Library. It was tested using a variety of experiments on each of the components of the process.  It shows promise for being able to replace or augment existing UGV perception systems in the future.

The U.S. military has increasingly turned to unmanned ground vehicles (UGVs) to assist in the most dull, dirty, and dangerous missions. Their presence on the battlefield is redefining how war is waged, expanding opportunities for reconnaissance and surveillance while minimizing soldier mortality. Robotic systems have gotten ever smaller, many now being man-packable. Soldiers may now carry, deploy, and control their own robotic assistant, many with limited formal training. Unfortunately, UGVs add significantly to a soldier's standard load of water, ammunition, armor, and supplies, making weight and portability top concerns. One way to ease the soldier burden is to adopt smartphones for use as robot operator control units (OCUs). Their small, lightweight frame combined with processing power and adaptable software backbone may enable intuitive controls on a device well-suited for other military missions.

OJHA, UNNATI. Delay Tolerant Behavior Control based Adaptive Bandwidth Allocation for Unmanned Ground Vehicles in a Network Control System. (Under the direction of Dr. Mo-Yuen Chow) The rapid growth of wired as well as wireless communication technology has led to the expansion of Network Control System (NCS) into Large Scale Distributed Network Control System. As the size of NCSs increase, important resources like bandwidth become scarce which degrade the performance of the NCS. It is necessary to manage the bandwidth in such a way that causes the least amount of degradation in the NCS performance. Many adaptive sampling methods have been suggested to meet the bandwidth constraint; however, efficiency of these methods decreases quickly as the amount of network delay increases. Delay in the network is unavoidable and stochastic thus it is important that a bandwidth management technique be robust to delays in order to maintain the performance of the NCS. This study proposes a Delay Tolerant Behavior Control (DTBC) based bandwidth allocation that is robust to network delays. This new method for bandwidth allocation in a system of Unmanned Ground Vehicles (UGVs) is inspired by the human driving behavior. In DTBC, bandwidth is allocated based on the UGV trajectory curvature, amount of network delay and the velocity of the UGV. A study is first presented to analyze the relationship between curvature and the UGV performance. The Behavior Control (BC) based bandwidth allocation algorithm, a predecessor of DTBC is then formulated. The BC algorithm is then expanded and improvised to formulate the DTBC algorithm. Both simulation and implementation results are presented at the end which confirms that DTBCs performance is superior to the existing algorithms especially in the presence of network delay.