<p>This master thesis includes a description of how a number of sensors are put together in a sensor platform. It also gives a description of how localization and mapping has been performed with data collected from the sensor platform. </p><p>Localization is a prerequisite for most of the tasks that can be requested from an autonomous mobile robot. In many situations the GPS signal is not available and hence an additional localization system is required. One approach is to apply localization based on landmarks extracted from the robots surrounding. Using e.g. a scanning range finding laser landmarks can be extracted from the robots surrounding. </p><p>Localization and mapping is performed with an extended kalman filter where the robots and the landmarks positions are represented as states. The positioning algorithm is finally tested in indoor environment using lines and line segments as landmarks. The landmarks are extracted from measurement data from the scanning range finding laser. This master thesis is a part of a larger project which purpose is to investigate the benefit of cooperation between several mobile robots.</p>

This thesis is divided in to two different parts. The first part includes examination of the measurementsystem of an industrial robot using a resolver sensor. The main focus is on methods for suppressing noise in the angularvelocity signal without increasing timedelay. Five different methods are investigated. Three of these are based on oversampling: burstsamplingmethod, meanvaluemethod and correlationmethod. The meanvaluemethod and the correlationmethod have given good results. The two other methods are: extended kalmanfiltering and computation of the angular velocity without using numeric method to compute the angle. Extended kalmanfilter gives the overall best result. In part two the control structure design of an industrial robot has been studied and how different sampling times and motorinertias affect the disturbance rejection and stability of the control loop. Different control structure designs have also been studied with the aim to suppress disturbances. Mainly H-infinity and GIMC designs have been compared with an ordinary PID controller. Here it can be shown that both the H-infinity and the GIMC controller yields good disturbance rejection. But both the methods lack in the robustness of model uncertainty.

This master thesis includes a description of how a number of sensors are put together in a sensor platform. It also gives a description of how localization and mapping has been performed with data collected from the sensor platform. Localization is a prerequisite for most of the tasks that can be requested from an autonomous mobile robot. In many situations the GPS signal is not available and hence an additional localization system is required. One approach is to apply localization based on landmarks extracted from the robots surrounding. Using e.g. a scanning range finding laser landmarks can be extracted from the robots surrounding. Localization and mapping is performed with an extended kalman filter where the robots and the landmarks positions are represented as states. The positioning algorithm is finally tested in indoor environment using lines and line segments as landmarks. The landmarks are extracted from measurement data from the scanning range finding laser. This master thesis is a part of a larger project which purpose is to investigate the benefit of cooperation between several mobile robots.

<p>This thesis is divided in to two different parts. The first part includes examination of the measurementsystem of an industrial robot using a resolver sensor. The main focus is on methods for suppressing noise in the angularvelocity signal without increasing timedelay. Five different methods are investigated. Three of these are based on oversampling: burstsamplingmethod, meanvaluemethod and correlationmethod. The meanvaluemethod and the correlationmethod have given good results. The two other methods are: extended kalmanfiltering and computation of the angular velocity without using numeric method to compute the angle. Extended kalmanfilter gives the overall best result. </p><p>In part two the control structure design of an industrial robot has been studied and how different sampling times and motorinertias affect the disturbance rejection and stability of the control loop. Different control structure designs have also been studied with the aim to suppress disturbances. Mainly H-infinity and GIMC designs have been compared with an ordinary PID controller. Here it can be shown that both the H-infinity and the GIMC controller yields good disturbance rejection. But both the methods lack in the robustness of model uncertainty.</p>