| Popis: |
In order to obtain an optimal motion of a mobile robot in different environmental structures, the motion strategy has to be adapted to the environment. For this, a classifier is needed that classifies the robots environment. Special driving methods have to be provided that lead to an optimal motion in the respective environment. In this paper, a survey of the architecture of motion control and its components is given. The sensor data of a laser range finder and sonar sensors are preprocessed by sensor data fusion and a growing technique. For typical office environments, a classifier is suggested for the classes corridor, wall, room, and door. For each environment class, a special driving method is described. Inside of rooms exact path following is performed. Inside of corridors the robot drives in the middle of the corridor or parallel to one wall at a defined distance. Wall following can also be performed, if only one wall is detected. Furthermore a method is presented that guides the robot through doors. The results of some test runs for the different methods are shown. |
