Dynamical Situation and Trajectory Discrimination by Means of Clustering and Accumulation of Raw Range Measurements

Abstract

This article focuses on the problem of identifying and discriminating situations and trajectories (as sequences of situations) in an autonomous mobile robot setup. The static identification level of situations as well as the dynamical level of trajectories are based on egocentric measurements only. Adaptation to a specific operating environment is performed in an exploration phase and continuously during operation. Descriptions and classifications are based on statistical entities of the operating environment (in the geometrical space and in the space of dynamics). The recognition is performed in the sense of emitting the same signals in similar situations or on similar trajectories. Neither a global position nor any other global geometrical description is created or employed by this approach.