Explorative experiments in autonomous robotics

Abstract

The debate on the experimental method, its role, its limits, and its possible applications has recently gained attention in autonomous robotics. If, from the one hand, classical experimental principles, such as repeatability and reproducibility, play as an inspiration for the development of good experimental practices in this research area, from the other hand, some recent analyses have evidenced that rigorous experimental approaches are not yet full part of the research habits in this community. In this paper, in order to give reason of a part of the current experimental practice in autonomous robotics that cannot be satisfactorily accommodated under the traditional concept of controlled experiment, we will advance the notion of explorative experiment. Explorative experiments in this context should be intended as a form of investigation carried out in the absence of a proper theory or theoretical background, where the control of the experimental factors cannot be fully managed from the beginning. We show that this notion arises from (and is supported by) the analysis of the experimental activities reported in a significant sample of papers that have been given awards at two of the largest and most impacting robotics research conferences.