Projekt: Navigation
Forscher: Sommer G. , Hailu G. , Bruske J.
One research project deals with robot navigation. In our group we are dealing for more then 8 years with this problem and analyse aspects of navigation within a behavior based system architecture.In its original sense the term navigation means to direct a ship to its destination (from Latin navis ship and agere to drive). Navigation means to answer three questions:
Where am I?, Where are other places with respect to me? and How do I get to other places from here?
In the context of robot navigation, the robot's sensory input must be used to answer the above questions. Therefor the following definition for navigation is also common:
A movable object has the task to move from its current position to a goal, on the base of fragmentary information and in consideration of boundary conditions.
The focus of our work is multilateral. On the one hand we deal with ultra-sonic based robot navigation, e.g. by the use of evidence grids for map building or we deal with vision based navigation for self-localization. The image shows our mobile robot within a openGL-simulation.
We use a B21-mobile robot which is equipped with a Pan-Tilt/vergence camera unit, ultra sonic sensors and infra-red sensors.
In previous works we also used a LabMate robot.
|
Please klick on the left image for a visualization of the selflocalization problem: The robot performs movement commands which are inexact. Accumulating such errors leads to non-tolerable results. The non-transparent robot visualizes the real position of the robot in 3D space and the transparent robot visualizes the robots internal position. Therefore we are interested in algorithms which help the mobile robot to self-localize itself. |
The following links separate our research in different topics and aspects:
- Ultra-sonic based robot navigation
- Vision based robot navigation
- Evolutionary robotics in navigation
