Online Detection of Instability for Robust Teamwork in Humanoid Soccer Robots

摘要

Abstract — Humanoid robots are increasingly being used in several domains as a result of the development of sophisticated algorithms for challenges related to vision, locomotion and team coordination. RoboCup is one such research initiative with the stated goal of creating a team of humanoid soccer robots that can beat the human soccer champion team on an outdoor soccer field by the year 2050 [1]. In order to achieve this goal, it is essential to enable a team of humanoid robots to collaborate effectively. In the Standard Platform League (SPL) of RoboCup [2], the instabilities induced by collisions, slippages and falls pose a major challenge to effective collaboration and provide a significant advantage to the opposing team. This paper presents an approach that enables a humanoid robot to autonomously learn a model based on sensory inputs to detect instabilities. The detected instabilities are communicated to teammates and incorporated in the team coordination strategy, resulting in a quick re-organization of the roles of the robots and more effective team coordination. All algorithms are implemented and tested on the Aldebaran Nao robot platform [3] in the robot soccer domain. I.