Vision-Based Detection and Pursuit of Humans with a Mobile Service Robot

Most application scenarios for autonomous mobile service robots require an intelligent and easy to use man-machine-interface. One of the basic capabilities in this context is the recognition and, in order to determine which user currently interacts with the robot, the identification of human users. Rather than performing recognition passively, integration into the current overall task is of great importance for a mobile system. Therefore, this research project covers both the selection and implementation of appropriate algorithms for person detection and tracking on a mobile platform and the combination with a navigation task in the form of a pursuit behaviour.

The following figure shows the results of the detection process for an image sequence:

 


Detection of human faces (yellow frame)

To illustrate the behaviour of pursuit, the path of a person walking along marking lines on the floor and the path of the pursuing robot are depicted in the following figure:


path of the pursuing robot