This tutorial will provide hands-on practical knowledge about how to perform reproducible robotics research and report it.
Reproducibility of results is a core characteristic of the scientific method. However, in Robotics and AI research the replicability and reproducibility of results and their objective evaluation and comparison are usually lacking. Defining comparable metrics and identifying goal similarity across various domains is still poorly understood. In the past determining the information required to enable replication of results has been the subject of extensive discussion. This situation has impaired and still impairs both research progress and technology transfer.
Luckily Reproducible Research in Robotics is now possible, thanks to the R-article process in IEEE Robotics and Automation Magazine. Since 2017 the IEEE Robotics & Automation Magazine solicits R-articles (i.e., papers that report experiments aiming to be fully reproducible), short articles reporting on the replication of r-Article results, and author short article replies. All papers are peer reviewed. The outcome is a two stage very high-quality review process. The first stage is provided by the ordinary rigorous review process of a top-tier publishing venue. The second stage will be the replication of the experiments by the community (what is usually the core of the scientific method). Several R-Articles have been already published and more are expected, so that it is now possible to share best practices in reproducible robotics research.