Probabilistic localization and mapping for mobile robots

The essential key capabilities for a mobile robot are to determine where it is located and gather an idea of its surroundings. For precise self localization several sensors are needed as due to noisy measurements no single sensor is sufficient. The data from the sensors is fused to a combined estimate resulting in a more accurate localization. As in most situations positioning sensors like odometry and GPS alone are still insufficient, for example in case of collision avoidance, it is preferable to incorporate exterozeptive sensors as well. Furthermore it is possible to use these sensors to localize the robot in a map. If a map of an area is unavailable, the robot has to build it while exploring the environment. This exploration and mapping is a very challenging problem because of noise in the sensor measurements and approaches to solve this so called simultaneous localization and mapping (SLAM) problem exist.