This video shows a very early prototype of the SPENCER robot guiding a group of travellers. The robot's task here is to drop off the group at a pre-defined goal. In the real use-case at an airport, the robot will pick up passengers with short connection times at their gate and guide them over distances of 400-800 meters.
This demo involves mapping & localization, people and group detection & tracking in 2D laser range data, task planning and motion planning. All modules are integrated via the Robot Operating System (ROS).
The video shows work in progress and was recorded during integration week II in October 2014. The people and group tracking system which is used here is a placeholder component based upon simple nearest-neighbor data association. In our current research, we are working on the integration of a multi-model multi-hypothesis group tracker with explicit tracking of group merge and split events. We will also integrate different RGB-D detectors for more robust person tracking.
