Nikos Tsakarakis, IIT, Italy
Zhibin Li, IIT, Italy
Achieving stable, efficient and versatile locomotion is one of the main goals of humanoid robots. In this context, the need for reliable benchmarks of bipedal locomotion is becoming more and more relevant, in order to assess the robotic behavior on a quantitative basis and to compare it with the human counterpart.
Such benchmarks or performance indicators are important in order to judge when a robot motion is “human-like” which is an often stated but never strictly defined goal. They are also important in order to compare the abilities of different robots to each other or to measure progress of an individual robot due to an implementation of new control methods or an exchange or parts. Despite this importance, no consensus has been achieved yet about the set of benchmarks that should be adopted.
This workshop aims to involve the international community interested in bipedal walking in a thorough discussion about the key features of human-like locomotion, with a specific focus on the assessment methods and protocols that can be easily applied across different laboratory settings. The aim of this workshop is to really WORK together and to come up, at the end of the day, with a preliminary, but specific proposal for benchmarks for humanoid walking.
In order to achieve this objective, the organization of this workshop will deviate from the usual structure of IEEE conference workshops based on plenary talks and discussions. After several very short challenge talks, the workshop will essentially be structured in small discussion groups, in order to promote real debate and reasoning, which possibly result in a preliminary proposal of benchmarks to be proposed to the international forum. The discussion groups will be focusing on different aspects related to functional performance and human-likeness, such as stability, efficiency, dynamic similarity, passive dynamics, etc. The discussion groups will then present their results to all workshop participants.
In order to be able to be really productive, this workshop will be well prepared via the newly established benchmarking mailing list and this webpage; and abstracts corresponding to the challenge talks will already be made available to the prospective audience in advance.
We strongly encourage the participation of all those researchers interested in sharing their ideas and willing to find common solutions.
More information available at: http://orb.iwr.uni-heidelberg.de/koroibot/?page_id=492