Oscillation analysis in behavior-based robot architectures

Abstract

This paper presents a method for detecting oscillations in behavior-based robot control networks. Two aspects are considered. On the one hand, the detection of oscillations inside single behavior modules is based on analyzing the signal in the frequency domain using the Fast Fourier Transformation (fft). On the other hand, tracing oscillations through the behavior network helps to evaluate its propagation and to find its root cause. Results of the oscillation analysis are presented using appropriate visualization techniques. The suitability of the proposed approach is shown by an indoor application.