This thesis presents the use of constraint programming for scheduling tasks in time triggered real time systems undergoing mode changes. The scheduling problem is initially decomposed into schedules for individual modes, and then the schedules generated for each operational mode are processed to maintain timing constraints across mode changes. These mode changes are allowed to occur at particular time instants specified by the designer. A mode change task, in effect, acts as a placeholder for these time instants and different heuristics are proposed to set the period of the mode change tasks. A mode change schedule, which is the new schedule that needs to be executed when a mode change occurs, can be subsequently generated. Constraint programming has previously been used to solve scheduling problems, but until now there have been few attempts to solve the mode change problem using this technique. An example of schedule generation with mode changes in an aircraft control system is shown.

Ionizing radiation poses a serious threat to semiconductor integrated circuits that are required to operate reliably in radiation-intensive environments, for example, circuits used in space electronics. Numerical modeling of radiation-induced defect formation in the Si-SiO2 system is finding increasing application in the design of radiation-resistant electronics. To date, several numerical hole-trapping simulators have been developed and applied to radiation-induced leakage problems. However, few attempts have been made to model the kinetics of interface trap formation. This dissertation presents a novel solution to this problem, specifically, a coupled model for electron, hole, and proton transport in Silicon Dioxide suitable for transient device simulation-based prediction of the buildup of both major types radiation-induced defects: trapped oxide charge and interface traps. This model provides two fundamental firsts in physically-based radiation effects simulation: (1) the representation of hole-trapping-induced proton release in a self-consistent system of electron, hole, and proton continuity equations, and (2) the application of a continuity equation-based model for dispersive proton transport. Essential features of hydrogen-mediated interface trap formation are demonstrated in a series of pulsed exposure/switched bias simulations.

In order to localize itself within an environment as well as navigate through an environment a mobile robot must have some idea of the layout of the environment. One useful method for storing relative information concerning the layout of an environment is through a process known as cognitive mapping. Richard Dagan defines cognitive mapping as a process composed of a series of psychological transformations by which an individual acquires, codes, stores, recalls, and decodes information about the relative locations and attributes of phenomena in their everyday spatial environment. In order to create such a cognitive map, it is important that the robot have the ability to sense the environment around it, as well as be able to extract useful information about the size, position, and orientation of features within the environment. This thesis will discuss an easy method to derive 3-D information from an environment utilizing a 2-D sensor (the highly popular SICK laser scanner). In addition, methods for determining size, position, and orientation information about features in the environment will be discussed. Combining several 3-D laser scans for the generation of a cognitive map will be discussed. Finally, the results derived from testing the cognitive map for localization and navigation purposes will be discussed.

<p>Engineering design is a process of tradeoff analysis. Design decisions are evaluated, often involving quantitative comparisons between relevant metrics of goodness. Design space exploration formalizes the process of tradeoff evaluation through a mathematical representation of design decisions and outcomes in a compositional design model. The set of all possible outcomes of tradeoff evaluation is called a design space. The quantitative evaluation of metrics defined over the space is referred to as design space exploration. The goal of design space exploration is to determine, through the evaluation of design metrics, a design composition or small set of compositions which meet a set of formally specified design criteria. <p>Design space exploration has arisen in the field of embedded system design, where complexities imposed by the physical environment of a system complicate the design process. The formal representations offered by most existing design space exploration tools are limited in scope to a particular problem or problem domain. Attempts to generalize existing exploration algorithms over a broad class of design space problems reveal scalability limitations. <p>This dissertation describes a hybrid approach to design space exploration, embodied in the DesertFD toolset. DesertFD offers a finite domain constraint implementation of a domain-independent formal design space model, together with pruning algorithms facilitating the application of OCL constraints to the design space. The tool offers an expressive language for specifying complex formulas for modeling evaluation metrics over the space, and provides an interpretation of the language into a finite domain constraint specification. DesertFD integrates the finite domain constraint based pruning with an existing symbolic pruning tool based on Ordered Binary Decision Diagrams, facilitating a hybrid approach to design space exploration.

In the past few decades unprecedented advances have been made in the field of medical imaging. Various imaging technologies such as Computed Tomography, Magnetic Resonance Imaging, etc. have emerged to assist the visualization of internal structures in the body. These along with the different image processing tools help in diagnosis and detection of diseases. Intensity Modulated Radiation Therapy (IMRT) is a recently developed and highly effective method for destroying cancerous cells with minimal effect on the other body structures of the patient. It relies on accurate delineation of the structures to be irradiated and those to be spared. Currently the delineation is being done manually, which is very time consuming. We have therefore proposed an atlas-based automatic segmentation method which will significantly reduce the interaction time during radiotherapy planning. In this thesis, the main focus is on improving the results obtained by the atlas-based segmentation method. Three methods have been implemented namely CT-MR fusion method, Mesh deformations and Classifier combination method. We have employed two methods for validating the automatically generated results. This is done by comparing the automatic masks and contours with the manual ground truth segmentation. Finally, the different methods have been compared and the feasibility of automatic delineation has been discussed.

Object recognition and learning algorithms are huge areas of robotics research with many different methods in use by various researchers. A common result of using complex recognition methods is the loss of meaning (for humans) in the subsequent processing of the data. When programs incorrectly identify objects, the reason why is often lost in the data analysis. If the researchers can understand what the robot sees, they are better able to develop a system that has limited image understanding. Fuzzy models in object recognition are one of the better methods for achieving such a learning system. Our desire is to develop a system that is quickly and easily trained, a system that can relate the decision of objects through the feature vector (vector of measured characteristics about the object), and a system that is relatively simple in its calculation of results. The research was applied in conjunction with an experiment done by the Psychology Department. This system was applied to recorded videos of tasks done by their subjects. The system proves to be quite effective in object recognition and provides many options for more advanced data processing.

The research in this thesis focuses on two problems related to the Sensory Ego-Sphere (SES), a short-term memory structure for a robot: (1) the mapping to the SES of high-resolution sensory information in the form of imagery, and (2) the concurrent processing of visual attention. Neither problem had been studied previously. The SES coordinates sensory information for further processing and thereby acts as an interface between sensing and cognition. It is an egocentric, spherical mapping of the robot's locale. This research is based on previous work in the areas of multi-modal sensing, sensory-motor coordination, and attention. The paper describes a procedure to composite on the SES an image sequence taken by a camera head, a task that is complicated by significant overlap between successive images in the sequence. Two approaches to the problem of finding and ranking areas of visual interest are compared. One combines visual attention points in the overlapping images on the SES and the other computes attentional points directly on the composited visual scene. Computational structures for mapping imagery to the SES and managing it are described. The problem of attention in bio-vision is discussed as are some algorithms that mimic visual attention behaviors in humans.

This research measures how well supervised-reinforcement-learning techniques perform when applied to real-world tasks, managed as a discrete-event dynamic system (DEDS). Two types of experiments are tested. One tests the robots stability in implementing a task it has been taught. The other experiment includes obstacles blocking the path to the goal and measures the robots flexibility. The supervisor consists of human-guided remote-controlled runs through the navigation task and acts as a teacher for the initial stages of reinforcement learning. Experimental analysis is based on measurements of average time to reach the goal and the number of failed states encountered during a trial of episodes.

Robot Navigation is a large component of current robotics research. In this paper, a method to implement the go-to-goal aspect of navigation is discussed. Specifically, I will first discuss a method called Egocentric Navigation, which was developed at Vanderbilt Universitys Center for Intelligent Systems. Then improvements toward making the system more accurate and efficient will be proposed. Finally, the experimental and simulation results of this new Egocentric Navigational system follow. In total, this paper will lay out a complete system for implementation on a robot which will accomplish navigation to goal in a landmark-based egocentric manner.

Hybrid System Based Design for the Coordination and Control of Multiple Autonomous Vehicles <p> Charles A. Clifton <p> Thesis under the direction of Professor Takkuen John Koo <p> In recent years, the use of unmanned aerial vehicles (UAVs) has gained considerable attention for applications where manned operation is considered dangerous or infeasible. As the number of UAVs in operation rises, it will become necessary to coordinate these vehicles. It can be shown that a real-time system can be modeled using a hybrid automaton provided that certain guarantees can be made about the temporal properties. By using the hybrid automata to model the system composed of a multi-modal dispatcher and waypoint/motion controller in addition to a real-time UAV controller, we show that the hybrid system can bisimulate a timed automata model created using a tool called UppAal, which can verify specifications about a given system. We thereby coordinate multi-robot movement while ensuring that certain constraints have not been violated. <p> In this thesis, we present a methodology for developing autonomous vehicle controllers using a model-based approach and a hybrid automata to represent the control system, which features both linear state feedback control and nonlinear control. We derive a linear model of the physical system by performing system identification and employ a Kalman Filter to obtain state estimates for feedback purposes. Furthermore, we explain how this can be constructed from a system containing both discrete-time linear and continuous-time nonlinear subsystems. In addition, we present the Vanderbilt Embedded Computing Platform for Autonomous Vehicles (VECPAV), an end-to-end design platform for the rapid development and deployment of control and motion planning solutions for autonomous vehicles. The automated development platform greatly speeds the controller and system development and deployment phases by reducing the programming and compilation burden on the lab researchers, and eliminating the risks associated with translating code manually.