Tutorial on
Practical Reproducibility
of Robotics Source Code
ICAR2023 Half-day Tutorial
Tuesday December 5th, from 9am to 1pm
In the daily practice of a robotics researcher, reusing the software of other labs for its integration in a new system or replicating their results is a real challenge. This uncomfortable situation does not only make very difficult the cumulative process of research, it also severely impairs technology transfer and industrial exploitation. The aim of this tutorial is to showcase concrete software practices and guidelines for making reproducible open source code in robotics and AI. This event will provide necessary information, knowledge and motivation to allow the community to transition to a reproducible software development process where seamless integration of the components of intelligent robotics and automation systems will be mainstream practice. We will review the past and present situation and focus on the prospective for the future, analyzing which test beds are currently available and for which specific field, what is missing and what, as a community, we need to provide for having more reproducible code in robotics research. We will present the evolution of software languages (C++, Python), libraries (OpenCV, ROS) and machine learning toolkits (TensorFlow, CUDA), and we will illustrate the problems of broken dependencies with examples of open source code repositories from actual conference and journal papers.
Organizers
Enric Cervera, RobInLab, Universitat Jaume-I
Fabio Bonsignorio, FER University of Zagreb
Angel P. del Pobil, RobInlab, Universitat Jaume-I
Dr. Enric Cervera is an active researcher in robotics and AI with a solid background in computer science and software engineering. He has recently investigated the problems of code reproducibility, and developed practical solutions and good practices for reusing robotics software.
Prof. Fabio Bonsignorio is founder and CEO of Heron Robots. He is ERA Chair in AI for Robotics at FER, University of Zagreb. He has been until 2019 a visiting professor at the Scuola Superiore Sant'Anna (Pisa, Italy). He is the coordinator of the euRobotics Topic Group on Experiment Replication, Benchmarking, Challenges and Competitions and is co-chair of the IEEE TC-Pebras.
Angel P. del Pobil, Professor of Computer Science and Artificial Intelligence (Engineering and Computer Science Deparment), is founder and Director of the Robotic Intelligence Laboratory at Jaume I University.
Program
09.00 Reproducible research in intelligent robotics: state of the art
Prof. Fabio Bonsignorio, FER U. of Zagreb
09.30 Reproducible articles with the H2Arm Open Platform
Dr. Enrica Zereik, CNR-INM, Head of Heron@CNR JoinLab
10.00 Software development in robotics and artificial intelligence
Dr. Enric Cervera, Robotic Intelligence Lab, UJI
10.30 Coffee break
11.00 Case studies: from a conference paper to a reproducible open source code repository
13.00 End of tutorial