Looking for concepts: unsupervised map construction with unknown sensor configuration

The navigation method presented allows a robot to find its position and its route without prior information about its environment. What is going to be used as landmark is initially unknown, and no specific preprocessing of the sensor signals is done. The robot nevertheless extracts meaningful information from its environment and uses it to build a map. Both landmark definitions and map are continuously adapted. Working without predefined categorization takes full advantage of the sensor abilities. The result shows that a robot can reliably recognize self-defined landmarks suitable for its sensor capability and create a map of its environment.