|
||
|
Mobile robots operating in populated environments should be able to perceive and react to the activities of people in their surrounding. Knowledge about the behavior of people within the environment enables the robot to improve its navigation behavior. For example, this ability allows a robot to adapt its velocity to the speed of people in the environment and to improve its collision avoidance behavior in situations in which the trajectory of the robot crosses the path of humans. Over the last years we have developed methods for tracking multiple moving objects with a mobile robot. These methods use the robot's sensors and motion models of the objects being tracked in order to estimate their position and velocities. The estimation process is based on efficient particle filtering techniques, which are able to cope with non-linear object dynamics as well as partial occlusion of objects. With our method we can keep track of the motions of several people in the vicinity of a moving robot and we are able to determine the location or state of other objects like doors and pieces of furniture. Future research focuses on hierarchical extensions of these methods which will enable mobile robots to assess the dynamic situation they are in, including the interactions between objects and the robot. This information will than be used to optimize the robots navigation plans. |