Can we Reproduce it? Toward the Implementation of good Experimental Methodology in Interdisciplinary Robotics Research

Lier F., Lücking P., de Leeuw J., Wachsmuth S., Šabanović S., Goldstone R. L.  (2017) Can we Reproduce it? Toward the Implementation of good Experimental Methodology in Interdisciplinary Robotics Research. In: Bonsignorio FP, ed. Can we Reproduce it? Toward the Implementation of good Experimental Methodology in Interdisciplinary Robotics Research. Proceedings of IEEE International Conference on Robotics and Automation (ICRA), IEEE Xplore. IEEE International Conference on Robotics and Automation (ICRA), Singapore: IEEE.

The insufficient level of reproducibility of published experimental results has been identified as a core issue in the field of robotics in recent years. Why is that? First of all, robotics focuses on the abstract concept of computation and the creation of technological artifacts, i.e., software that implements these concepts. Hence, before actually reproducing an experiment, the subject of investigation must be artificially created, which is non-trivial given the inherent complexity [5]. Second, robotics experiments usually include expensive and often customized hardware setups (robots), that are difficult to operate for non-experts. Finally, there is no agreed upon set of methods in order to setup, execute, or (re-)conduct an experiment.

To this end, we introduce an interdisciplinary and geographically distributed collaboration project that aims at implementing good experimental methodology in interdisciplinary robotics research with respect to: a) reproducibility of required technical artifacts, b) explicit and comprehensible experiment design, c) repeatable/reproducible experiment execution, and d) reproducible evaluation of obtained experiment data. The ultimate goal of this collaboration is to reproduce the same experiment in two different laboratories using the same systematic approach which is presented in this work.

Download PDF of article