Localization and object detection for outdoor mobile robots with 2-D laser sensor

Applications of outdoor mobile robots can be found in various diversified fields including defense, agriculture, mining and space exploration. To carry out their tasks successfully, the robots have to locate the pose in the environment and detect the objects around them. This thesis presents the development and implementation of localization enhancement and object detection using a laser range sensor and the map service used for localization and object detection intentions.;The problem of gate-like object recognizing and crossing is described at first. The novel concept of signature is proposed and applied. The algorithm based on the signatures from the special objects is developed. A "gating controller" is designed by extending the classical parking controller. The simulations and experiments demonstrate the mobile robot approaching and crossing the gate-like object using the developed algorithms. For the generalization of the concept of signature from the special object, a data association algorithm based on geometry features is developed. Using this algorithm, the data from a 2-D laser range sensor is compared with a reference scan from the simulator of the physical laser sensor to locate the pose of the robot. Based on the matching algorithm, two JAUS compatible components, Range Based Pose and Object Tracking, are developed and integrated into the mobile robot system. Simulation and experiment about localization and object tracking tasks are discussed.