Design of an autonomous office guide

In this master thesis a guiding software for an autonomous robot was developed.The purpose was to enable a preexisting robot to guide visitors to the right office space inside of Aros Electronics.The guiding software have four major parts; path planning, path following, collision avoidance and localization.The A* algorithm is used on a map of the office to get the path to the destination.The path following is done by driving the robot towards a waypoint located on the path ahead of the robot.The collision avoidance checks that the path is clear and if not the robot is not allowed to continue.Two ways of localization is tested in combination with odometry.First an Xbox kinect camera is used to spot landmarks and the distance and angle to this landmark is used in a probabilistic method called particle filter to estimate the pose.Secondly a wall following method is developed where ultrasonic proximity sensors are used to sense the walls of the office.Both localization methods greatly improves the pose estimation compared to using only odometry but the wall following method shows better performance compared to particle filter method.The main reason for this is that the update frequency is much higher.