No description available.

One of the main problems in the design of mobile robots is the development of creating smart integrated information systems. These systems may include different type of sensors. Usually a CeD vision system is a necessary part for these systems. This thesis considers the design of a fast mechanical scanning system for a CCD vision system and the synthesis of the optimal control for this system. The mathematical model of the transport subsystem for mobile robots subjected to an external disturbance is created. The correctness of this model is proved on the base of simulation and experimental results. Regression coefficients are calculated and an estimate of model accuracy is carried out. The approach of power calculation for the actuators for the fast mechanical scanning system is considered. Implementation of this approach extends the use of a general mathematical model of the transport subsystem of the mobile robot and as such considerably reduces the design time. The main features for estimation of external disturbances are determined. Limitations of implementation of some technical solutions for mobile robot sensors are defined according to an analysis conducted of different factors for external disturbances. Di~erent kinematic schemes of the scanning systems have been analysed in this thesis. Practical recommendations of the kinematic scheme used in scanning systems are given. The essential features of a kinematic scheme for the fast mechanical scanning system have been developed and verified. A method for the solution of the inverse kinematics of a 3 degree of freedom scanning system in terms of velocities and in accelerations is presented. This method is utilised for formulating optimal control for the fast mechanical scanning system. 11 v ASlLY RUBrsov PHD· THESIS The algorithms of fast scanning have been produced for the different types of the sensors. The limitations for the practical realisations for these algorithms are considered in this thesis. The optimal control algorithms for the developed scanning system are produced. This control minimises the sum of instant powers of the scanning system actuators. A practical algorithm has been derived utilising the control scheme structure developed theoretically. The capability of this control algorithm has been proved by experimental study. Advantages of this developed control algorithm for 3 degree scanning system has been proven by experimentation.

No description available.

This thesis proposes and designs an autonomous transportation robot. It can provide elders and disabled people to cross the crosswalk safely with sensor information fusion for real time decision-making and control by fuzzy inference and the information of image, radar, and encoder etc. In this study, the vision system makes feedback of position offset, and declination angle through the fuzzy controller, the robot could modify the attitude error to cross the crosswalk completely. This thesis considers obstacles in the forward path. Using the vision feedback information to design the fuzzy controller, it can supply obstacle avoidance to make the transportation smoothly. Besides, we also address the robot guide at a crosswalk interface point. The detection of special purpose road for the blinds on the vision system can guide the robot to make a right turn or left turn to the next intersection. In cooperate with the intersection-agent system for collision avoidance, the robot could pass through the next crosswalk safely in order to finish the whole safely intersection system.

This dissertation work is describing the usage of the software tool Computer Vision System Toolbox to create applications in computer vision. At the beginning of the work is performed background research of image scanning and its representation by using colour models. It is followed by a description of epipolar geometry and lastly is stated a description of the Computer Vision System Toolbox. In the next section of the work we deal with setting of used Basler cameras and processing of the scanned image. The following is a description how to create applications for object detection and after this description, we get to know applications for creation of depth maps area.

No description available.

This thesis describes the development of a software system for color sorting hardwood edge-glued panel parts. Conceptually, this system can be broken down into three separate processing steps. The first step is to segment color images of each of the two part faces into background and part. The second step involves extracting color information from each region labeled part and using this information to classify each part face as one of a pre-selected number of color classes plus an out class. The third step involves using the two face labels and some distance information to determine which part face is the better to use in the face of an edge-glued panel. Since a part face is illuminated while the background is not, the segmentation into background and part can be done using very simple computational methods. The color classification component of this system is based on the Trichromatic Color Theory. It uses an estimate of a part's 3-dimension (3-D) color probability function, P, to characterize the surface color of the part. Each color class is also represented by an estimate of the 3-D color probability function that describes the permissible distribution of colors within this color class. Let P_omega_i denote the estimated probability function for color class omega_i. Classification is accomplished by finding the color difference between the estimated color probability function for the part and each of the estimated 3-D color probability functions that represent the color classes. The distance function used is the sum of the absolute values of the differences between the elements of the estimated probability function for a class and the estimated probability function of the part. The sample is given the label of the color class to which it is closest if this distance is less than some class specific threshold for that class. If the distance to the class to which the part is closest is larger than the threshold for that class, the part is called an out. This supervised classification procedure first requires one to select training samples from each of the color classes to be considered. These training samples are used to generate P_omega_i for each color class omega_i and to establish the value of the threshold T_i that is used to determine when a part is an out. To aid in determining which part face is better to use in making a panel, the system allows one to prioritize the various color classes so that one or more color classes can have the same priority. Using these priorities, labels for each of the part faces, and the distance from each of the part faces' estimated probability functions to the estimated probability function of the class to which each face was assigned, the decision logic selects which is the ``better'' face. If the two part faces are assigned to color classes that have different priorities, the part face assigned to the color class with higher priority is chosen as the better face. If the two part faces have been assigned to the same color class or to two different classes having the same priority, the part face that is closest to the estimated probability function of the color class to which it has been assigned is chosen to be the better face. Finally, if both faces are labeled out, the part becomes an out part. This software system has been implemented on a prototype machine vision system that has undergone several months of in-plant testing. To date the system has only been tested on one type of material, southern red oak, with which it has proven itself capable of significantly out performing humans in creating high-quality edge-glued panels. Since southern red oak has significantly more color variation than any other hardwood type or species, it is believed that this system will work very well on any hardwood material. / Master of Science

Autonomy is the most enabling and the least developed robot capability. A mobile robot is autonomous if capable of independently attaining its objectives in unpredictable environment. This requires interaction with the environment by sensing, assessing, and responding to events. Such interaction has not been achieved. The core problem consists in limited understanding of robot autonomy and its aspects, and is exacerbated by the limited resources available in a small autonomous mobile robot such as energy, information, and space. This thesis describes an efficient biomorphic visual capability that can provide purposeful interaction with environment for a small autonomous mobile robot. The method used for achieving this capability comprises synthesis of an integral paradigm of a purposeful autonomous mobile robot, formulation of requirements for the visual capability, and development of efficient algorithmic and technological solutions. The paradigm is a product of analysis of fundamental aspects of the problem, and the insights found in inherently autonomous biological organisms. Based on this paradigm, analysis of the biological vision and the available technological basis, and the state-of-the-art in vision algorithms, the requirements were formulated for a biomorphic visual capability that provides the situation awareness capability for a small autonomous mobile robot. The developed visual capability is comprised of a sensory and processing architecture, an integral set of motion vision algorithms, and a method for visual ranging of still objects that is based on them. These vision algorithms provide motion detection, fixation, and tracking functionality with low latency and computational complexity. High temporal resolution of CMOS imagers is exploited for reducing the logical complexity of image analysis, and consequently the computational complexity of the algorithms. The structure of the developed algorithms conforms to the arithmetic and memory resources available in a system on a programmable chip (SoPC), which allows complete confinement of the high-bandwidth datapath within a SoPC device and therefore high-speed operation by design. The algorithms proved to be functional, which validates the developed visual capability. The experiments confirm that high temporal resolution imaging simplifies image motion structure, and ultimately the design of the robot vision system.

Artificial intelligent (AI) systems today are booming and they are used to solve new tasks or improve the performance on existing ones. However, most AI systems work in a black-box fashion, which prevents the users from accessing the inner modules. This leads to two major problems: (i) users have no idea when the underlying system will fail and thus it could fail abruptly without any warning or explanation, and (ii) users' lack of proficiency about the system could fail pushing the AI progress to its state-of-the-art. In this work, we address these problems in the following directions. First, we develop a failure prediction system, acting as an input filter. It raises a flag when the system is likely to fail with the given input. Second, we develop a portfolio computer vision system. It is able to predict which of the candidate computer vision systems perform the best on the input. Both systems have the benefit of only looking at the inputs without running the underlying vision systems. Besides, they are applicable to any vision system. By equipped such systems on different applications, we confirm the improved performance. Finally, instead of identifying errors, we develop more interpretable AI systems, which reveal the inner modules directly. We take two tasks as examples, words semantic matching and Visual Question Answering (VQA). In VQA, we take binary questions on abstract scenes as the first stage, then we extend to all question types on real images. In both cases, we take attention as an important intermediate output. By explicitly forcing the systems to attend correct regions, we ensure the correctness in the systems. We build a neural network to directly learn the semantic matching, instead of using the relation similarity between words. Across all the above directions, we show that by diagnosing errors and making more interpretable systems, we are able to improve the performance in the current models. / Ph. D.

No description available.