Although the theory of Simultaneous Localization and Map Building (SLAM) is well developed, there are many challenges to overcome when incorporating vision sensors into SLAM systems. Visual sensors have different properties when compared to range finding sensors and therefore require different considerations. Existing vision-based SLAM algorithms extract point landmarks, which are required for SLAM algorithms such as the Kalman filter. Under this restriction, the types of image features that can be used are limited and the full advantages of vision not realized. This thesis examines the theoretical formulation of the SLAM problem and the characteristics of visual information in the SLAM domain. It also examines different representations of uncertainty, features and environments. It identifies the necessity to develop a suitable framework for vision-based SLAM systems and proposes a framework called VisionSLAM, which utilizes an appearance-based landmark representation and topological map structure to model metric relations between landmarks. A set of Haar feature filters are used to extract image structure statistics, which are robust against illumination changes, have good uniqueness property and can be computed in real time. The algorithm is able to resolve and correct false data associations and is robust against random correlation resulting from perceptual aliasing. The algorithm has been tested extensively in a natural outdoor environment.

Autonomous forklift handling systems is one of the most interesting research in the last decades. While research fields such as path planning and map building are taking the most significant work for other type of autonomous vehicles, detecting objects that need to move and picking it up becomes one of the most important research fields in autonomous forklifts field. We in this research had provided an algorithm for detecting paper reels accurate position in paper reels warehouses giving a map of the warehouse itself. Another algorithm is provided for giving the priority of papers that want to be picked up. Finally two algorithms for choosing the most appropriate direction for picking the target reel and for choosing the safest path to reach the target reel without damage it are provided. While working on the last two algorithms shows very nice results, building map for unknown stake of papers by accumulating maps over time still tricky. In the following pages we will go in detail by the steps that we followed to provide these algorithms started from giving an over view to the problem background and moving through the method that we used or we developed and ending by result and the conclusion that we got from this work.

A wearable robot should know its environment and its location in order to help its operator. Wearable robots are becoming more feasible with the development of more powerful and smaller computing devices and cameras. The main aim of this research is to build a wearable robot with a low cost stereo camera system which explores a room sized unknown environment online and automatically. To achieve 3D localization and map building for the wearable robot, a consistent visual-SLAM algorithm is implemented by using point features in the environment and Extended Kalman Filter for state estimation. The whole system includes camera models and calibration, feature extraction, depth measurement and Extended Kalman Filter algorithm. Moreover, a map management algorithm is developed. This algorithm keeps the number of features spatially uniform in the scene and adds new features when feature number decreases in a frame. Furthermore, a user-interface is presented so that the location of the camera,the features and the constructed map are visualized online. Most importantly, the system is conducted by a low-cost stereo system.

Unmanned air vehicles (UAVs) have become an important area of research. UAVs are used in many environments which may have previously unknown obstacles or sources of danger. This research addresses the problem of obstacle mapping and path planning while the UAV is in flight. Online obstacle mapping is achieved through the use of a multi-resolution map. As sensor information is received, a quadtree is built up to hold the information based upon the uncertainty associated with the measurement. Once a quadtree map of obstacles is built up, we desire online path re-planning to occur as quickly as possible. We introduce the idea of a quadtree rapidly-exploring random tree (RRT), which will be used as the online path re-planning algorithm. This approach implements a variable sized step instead of the fixed-step size usually used in the RRT algorithm. This variable step uses the structure of the quadtree to determine the step size. The step size grows larger or smaller based upon the size of the area represented by the quadtree it passes through. Finally this approach is tested in a simulation environment. The results show that the quadtree RRT requires fewer steps on average than a standard RRT to find a path through an area. It also has a smaller variance in the number of steps taken by the path planning algorithm in comparison to the standard RRT.

O problema de navegação de robôs móveis tem sido estudado ao longo de vários anos, com o objetivo de se construir um robô com elevado grau de autonomia. O aumento da autonomia de um robô móvel está relacionado com a capacidade de aquisição de informações e com a automatização de tarefas, tal como a construção de mapas de ambiente. Sistemas de visão são amplamente utilizados em tarefas de robôs autônomos devido a grande quantidade de informação contida em uma imagem. Além disso, sensores omnidirecionais catadióptricos permitem ainda a obtenção de informação visual em uma imagem de 360º, dispensando o movimento da câmera em direções de interesse para a tarefa do robô. Mapas de ambiente podem ser construídos para a implementação de estratégias de navegações mais autônomas. Nesse trabalho desenvolveu-se uma metodologia para a construção de mapas para navegação, os quais são a representação da geometria do ambiente. Contém a informação adquirida por um sensor catadióptrico omnidirecional estéreo, construído por uma câmera e um espelho hiperbólico. Para a construção de mapas, os processos de alinhamento, correspondência e integração, são efetuados utilizando-se métricas de diferença angular e de distância entre os pontos. A partir da fusão dos mapas locais cria-se um mapa global do ambiente. O processo aqui desenvolvido para a construção do mapa global permite a adequação de algoritmos de planejamento de trajetória, estimativa de espaço livre e auto-localização, de maneira a obter uma navegação autônoma. / The problem of mobile robot navigation has been studied for many years, aiming at build a robot with an high degree of autonomy. The increase in autonomy of a mobile robot is related to its capacity of acquisition of information and the “automation” of tasks, such as the environment map building. In this aspect vision has been widely used due to the great amount of information in an image. Besides that catadioptric omnidirectional sensors allow to get visual information in a 360o image, discharging the need of camera movement in directions of interest for the robot task. Environment maps may be built for an implementation of strategies of more autonomous navigations. In this work a methodology is developed for building maps for robot navigations, which are the representation of the environment geometry. The map contains the information received by a stereo omnidirectional catadioptric sensor built by a camera and a hyperbolic mirror. For the map building, the processes of alignment, registration and integration are performed using metric of angular difference and distance between the points. From the fusion of local maps a global map of the environment is created. The method developed in this work for global map building allows to be coupled with algorithms of path planning, self-location and free space estimation, so that autonomous robot navigation can be obtained.

O problema de navegação de robôs móveis tem sido estudado ao longo de vários anos, com o objetivo de se construir um robô com elevado grau de autonomia. O aumento da autonomia de um robô móvel está relacionado com a capacidade de aquisição de informações e com a automatização de tarefas, tal como a construção de mapas de ambiente. Sistemas de visão são amplamente utilizados em tarefas de robôs autônomos devido a grande quantidade de informação contida em uma imagem. Além disso, sensores omnidirecionais catadióptricos permitem ainda a obtenção de informação visual em uma imagem de 360º, dispensando o movimento da câmera em direções de interesse para a tarefa do robô. Mapas de ambiente podem ser construídos para a implementação de estratégias de navegações mais autônomas. Nesse trabalho desenvolveu-se uma metodologia para a construção de mapas para navegação, os quais são a representação da geometria do ambiente. Contém a informação adquirida por um sensor catadióptrico omnidirecional estéreo, construído por uma câmera e um espelho hiperbólico. Para a construção de mapas, os processos de alinhamento, correspondência e integração, são efetuados utilizando-se métricas de diferença angular e de distância entre os pontos. A partir da fusão dos mapas locais cria-se um mapa global do ambiente. O processo aqui desenvolvido para a construção do mapa global permite a adequação de algoritmos de planejamento de trajetória, estimativa de espaço livre e auto-localização, de maneira a obter uma navegação autônoma. / The problem of mobile robot navigation has been studied for many years, aiming at build a robot with an high degree of autonomy. The increase in autonomy of a mobile robot is related to its capacity of acquisition of information and the “automation" of tasks, such as the environment map building. In this aspect vision has been widely used due to the great amount of information in an image. Besides that catadioptric omnidirectional sensors allow to get visual information in a 360o image, discharging the need of camera movement in directions of interest for the robot task. Environment maps may be built for an implementation of strategies of more autonomous navigations. In this work a methodology is developed for building maps for robot navigations, which are the representation of the environment geometry. The map contains the information received by a stereo omnidirectional catadioptric sensor built by a camera and a hyperbolic mirror. For the map building, the processes of alignment, registration and integration are performed using metric of angular difference and distance between the points. From the fusion of local maps a global map of the environment is created. The method developed in this work for global map building allows to be coupled with algorithms of path planning, self-location and free space estimation, so that autonomous robot navigation can be obtained.

Labor shortage has prompted researchers to develop robot platforms for agriculture field scouting tasks. Sensor-based automatic topographic mapping and scouting algorithms for rough and large unstructured environments were presented. It involves moving an image sensor to collect terrain and other information and concomitantly construct a terrain map in the working field. In this work, a triangular mesh map was first used to represent the rough field surface and plan exploring strategies. A 3D image sensor model was used to simulate collection of field elevation information.<p>A two-stage exploring policy was used to plan the next best viewpoint by considering both the distance and elevation change in the cost function. A greedy exploration algorithm based on the energy cost function was developed; the energy cost function not only considers the traveling distance, but also includes energy required to change elevation and the rolling resistance of the terrain. An information-based exploration policy was developed to choose the next best viewpoint to maximise the information gain and minimize the energy consumption. In a partially known environment, the information gain was estimated by applying the ray tracing algorithm. The two-part scouting algorithm was developed to address the field sampling problem; the coverage algorithm identifies a reasonable coverage path to traverse sampling points, while the dynamic path planning algorithm determines an optimal path between two adjacent sampling points.<p>The developed algorithms were validated in two agricultural fields and three virtual fields by simulation. Greedy exploration policy, based on energy consumption outperformed other pattern methods in energy, time, and travel distance in the first 80% of the exploration task. The exploration strategy, which incorporated the energy consumption and the information gain with a ray tracing algorithm using a coarse map, showed an advantage over other policies in terms of the total energy consumption and the path length by at least 6%. For scouting algorithms, line sweeping methods require less energy and a shorter distance than the potential function method.

Labor shortage has prompted researchers to develop robot platforms for agriculture field scouting tasks. Sensor-based automatic topographic mapping and scouting algorithms for rough and large unstructured environments were presented. It involves moving an image sensor to collect terrain and other information and concomitantly construct a terrain map in the working field. In this work, a triangular mesh map was first used to represent the rough field surface and plan exploring strategies. A 3D image sensor model was used to simulate collection of field elevation information.<p>A two-stage exploring policy was used to plan the next best viewpoint by considering both the distance and elevation change in the cost function. A greedy exploration algorithm based on the energy cost function was developed; the energy cost function not only considers the traveling distance, but also includes energy required to change elevation and the rolling resistance of the terrain. An information-based exploration policy was developed to choose the next best viewpoint to maximise the information gain and minimize the energy consumption. In a partially known environment, the information gain was estimated by applying the ray tracing algorithm. The two-part scouting algorithm was developed to address the field sampling problem; the coverage algorithm identifies a reasonable coverage path to traverse sampling points, while the dynamic path planning algorithm determines an optimal path between two adjacent sampling points.<p>The developed algorithms were validated in two agricultural fields and three virtual fields by simulation. Greedy exploration policy, based on energy consumption outperformed other pattern methods in energy, time, and travel distance in the first 80% of the exploration task. The exploration strategy, which incorporated the energy consumption and the information gain with a ray tracing algorithm using a coarse map, showed an advantage over other policies in terms of the total energy consumption and the path length by at least 6%. For scouting algorithms, line sweeping methods require less energy and a shorter distance than the potential function method.

For an autonomous mobile robot, localization and map building are vital capabilities. The localization ability provides the robot location information, so the robot can navigate in the environment. On the other hand, the robot can interact with its environment using a model of the environment (map information) which is provided by map building mechanism. These two capabilities depends on each other and simultaneous operation of them is called SLAM (Simultaneous Localization and Map Building). While various sensors are used for this algorithm, vision-based approaches are relatively new and have attracted more interest in recent years. In this thesis work, a versatile Visual SLAM system is constructed and presented. In the core of this work is a vision-based simultaneous localization and map building algorithm which uses point features in the environment as visual landmarks and Extended Kalman Filter for state estimation. A detailed analysis of this algorithm is made including state estimation, feature extraction and data association steps. The algorithm is extended to be used for both stereo and single camera systems. The core of both algorithms is same and we mention the differences of both algorithms originated from the measurement dissimilarity. The algorithm is run also in different motion modes, namely predefined, manual and autonomous. Secondly, a map management strategy is developed especially for extended environments. When the robot runs the SLAM algorithm in large environments, the constructed map contains a great number of landmarks obviously. The efficiency algorithm takes part, when the total number of features exceeds a critical value for the system. In this case, the current map is rarefied without losing the geometrical distribution of the landmarks. Furthermore, a well-organized graphical user interface is implemented which enables the operator to select operational modes, change various parameters of the main SLAM algorithm and see the results of the SLAM operation both textually and graphically. Finally, a basic mission concept is defined in our system, in order to illustrate what robot can do using the outputs of the SLAM algorithm. All of these ideas mentioned are implemented in this thesis, experiments are conducted using a real robot and the analysis results are discussed by comparing the algorithm outputs with ground-truth measurements.