There are two commonly accepted paradigms for organizing intelligence in robotic vehicles, namely reactive and deliberative. A third, a hybrid paradigm called integrated planning and execution, is considered a combination of the original two. Although these paradigms are well known to researchers, there are few published examples directly comparing their application and performance on similar vehicles operating in identical environments. Virginia Tech's participation with two nearly identical vehicles in the DARPA Grand Challenge afforded a practical opportunity for such a case study. Both base vehicles were developed by modifying Club Car Pioneer XRT 1500 on-demand four wheel drive base platforms. Cliff was designed to use the reactive paradigm, while Rocky was designed to use the deliberative paradigm. Both vehicles were initially outfitted with sensor suites and computational capabilities commensurate with the paradigm being employed. The author of this thesis coordinated the activities of the two teams of undergraduate and graduate students who implemented the respective designs and software. Both vehicles proved capable of off-road navigation, including road following and obstacle avoidance in complex desert terrain. In the end, however, the reactive paradigm proved to be smoother and more reliable than the deliberative paradigm under the conditions of our testing. While both vehicles were extensively tested and compared using the competing paradigms, the team modified Rocky to use the more effective reactive paradigm for the Grand Challenge events. The deliberative case shows much promise for complex navigation, but added unnecessary complexity to desert road navigation. This case study, while necessarily limited in scope, may help to shed additional light on the tradeoffs and performance of competing approaches to machine intelligence. / Master of Science

This thesis describes the initial development of the Texas A&M Autonomous Ground Vehicle test platform and waypoint following software, including the associated controller design. The original goal of the team responsible for the development of the vehicle was to enter the DARPA Grand Challenge in October 2005. A 2004 Ford F150 4x4 pickup was chosen as the vehicle platform and was modified with a 6” suspension lift and 35” tires, as well as a commercial drive-by-wire system. The waypoint following software, the design of which is described in this thesis, is written in C and successfully drives the vehicle on a course defined by GPS waypoints at speeds up to 50 mph. It uses various heuristics to determine desired speeds and headings and uses control feedback to guide the vehicle towards these desired states. A vehicle dynamics simulator was also developed for software testing. Ultimately, this software will accept commands from advanced obstacle avoidance software so that the vehicle can navigate in true off-road terrain.

Previous methods for planning clothoid based continuous curvature paths aim at minimizing path length. However, minimal length paths are not always smooth, natural, and drivable. A method of generating clothoid-based trajectories is discussed using constructive polylines. The goal of the motion planner is to create a path for a large car-like vehicle in human driving environments. Thus, the trajectories generated by the motion planner must be smooth, drivable, and natural such that the vehicle can follow the planned path on human roadways. Several examples are shown of trajectories developed for a DARPA Urban Challenge vehicle and a method of testing the motion planner and the vehicle controller is described.

Master`s thesis is focused on mobile robotics and computer vision. There is briefly introduced a library of functions for image processing OpenCV. Then it deals with image processing and navigation of mobile robots using image data. There are described segmentation methods and methods for navigating through feature points.

Global food systems are facing a grand challenge due to daunting food security targets, risk vulnerabilities (e.g., COVID-19 crisis), internal unsustainability regarding resource use and contribution to environmental degradation (e.g., GHG emissions, land degradation and biodiversity loss). Such a situation demands a transformation of the global food system towards more local, alternative, community, and sustainable food systems. Past studies of the food system have shown that major advances toward socio-environmental sustainability can be achieved through the application of technology and the push for innovation. Therefore, since water is at the center of all food system challenges this study focused on climate-resilient water innovation, which includes using hydroponics, aeroponics or aquaponics technologies within a Controlled Environment Agriculture (CEA) to bring about this change. This qualitative study was undertaken to uncover the individual, institutional, technological, and farm/organizational factors that influenced the adoption of such climate-resilient water technologies. Additionally, it also involved investigating the individual construals and their relation to sustainable water innovation adoption. The study was able to discover new constructs, namely- sustainability ethos, individual construals, land/soil characteristics, location of urban farm, farm size bias, inter-provincial variation, complex systemic technologies, localization, and all-year-round production, that act as barriers and facilitators to sustainable water innovation adoption. Finally, developed a sustainable water innovation adoption model using an inductive and deductive approach that can be used by technology providers, the government and policy institutions for insightful decision-making with respect to water innovation in the Canadian agri-food sector.

In this thesis, the expression variation problem in two-dimensional (2D) and three-dimensional (3D) face recognition is tackled. While discriminant analysis (DA) methods are effective solutions for recognizing expression-variant 2D face images, they are not directly applicable when only a single sample image per subject is available. This problem is addressed in this thesis by introducing expression subspaces which can be used for synthesizing new expression images from subjects with only one sample image. It is proposed that by augmenting a generic training set with the gallery and their synthesized new expression images, and then training DA methods using this new set, the face recognition performance can be significantly improved. An important advantage of the proposed method is its simplicity; the expression of an image is transformed simply by projecting it into another subspace. The above proposed solution can also be used in general pattern recognition applications. The above method can also be used in 3D face recognition where expression variation is a more serious issue. However, DA methods cannot be readily applied to 3D faces because of the lack of a proper alignment method for 3D faces. To solve this issue, a method is proposed for sampling the points of the face that correspond to the same facial features across all faces, denoted as the closest-normal points (CNPs). It is shown that the performance of the linear discriminant analysis (LDA) method, applied to such an aligned representation of 3D faces, is significantly better than the performance of the state-of-the-art methods which, rely on one-by-one registration of the probe faces to every gallery face. Furthermore, as an important finding, it is shown that the surface normal vectors of the face provide a higher level of discriminatory information rather than the coordinates of the points. In addition, the expression subspace approach is used for the recognition of 3D faces from single sample. By constructing expression subspaces from the surface normal vectors at the CNPs, the surface normal vectors of a 3D face with single sample can be synthesized under other expressions. As a result, by improving the estimation of the within-class scatter matrix using the synthesized samples, a significant improvement in the recognition performance is achieved.

In this thesis, the expression variation problem in two-dimensional (2D) and three-dimensional (3D) face recognition is tackled. While discriminant analysis (DA) methods are effective solutions for recognizing expression-variant 2D face images, they are not directly applicable when only a single sample image per subject is available. This problem is addressed in this thesis by introducing expression subspaces which can be used for synthesizing new expression images from subjects with only one sample image. It is proposed that by augmenting a generic training set with the gallery and their synthesized new expression images, and then training DA methods using this new set, the face recognition performance can be significantly improved. An important advantage of the proposed method is its simplicity; the expression of an image is transformed simply by projecting it into another subspace. The above proposed solution can also be used in general pattern recognition applications. The above method can also be used in 3D face recognition where expression variation is a more serious issue. However, DA methods cannot be readily applied to 3D faces because of the lack of a proper alignment method for 3D faces. To solve this issue, a method is proposed for sampling the points of the face that correspond to the same facial features across all faces, denoted as the closest-normal points (CNPs). It is shown that the performance of the linear discriminant analysis (LDA) method, applied to such an aligned representation of 3D faces, is significantly better than the performance of the state-of-the-art methods which, rely on one-by-one registration of the probe faces to every gallery face. Furthermore, as an important finding, it is shown that the surface normal vectors of the face provide a higher level of discriminatory information rather than the coordinates of the points. In addition, the expression subspace approach is used for the recognition of 3D faces from single sample. By constructing expression subspaces from the surface normal vectors at the CNPs, the surface normal vectors of a 3D face with single sample can be synthesized under other expressions. As a result, by improving the estimation of the within-class scatter matrix using the synthesized samples, a significant improvement in the recognition performance is achieved.

The main task of this master’s thesis is to design circuits for feedback control of the main drives Minidarpa robot. It contains the description of power-driven mobile robot control theory and the DC motor. The second part deals with the design options of the control module and its mechanical design. The last part of the proposal made cascade speed control with current loop by using optimal module and the symmetric optimum methods.