Challenges of Multi-Robot World Modelling in Dynamic and Adversarial Domains

摘要

In all of the RoboCup soccer leagues, teams of robots compete to score goals in the presence of opponent robots. We focus on the RoboCup Standard Platform League (SPL), in which the robot platform is the same for all the competing teams, and the robots are fully autonomous with onboard directional perception, computation, action, and wireless communication among them. We address the problem of each robot building a model of the world in real-time, given a combination of its own limited sensing, known models of actuation, and the communicated information from its teammates. Such multi-robot world modelling is quite complicated due to the limited perception and the tight coupling between behaviors, sensing, localization, and communication. We describe the world model problem for the RoboCup SPL in detail, in particular with respect to the real-world constraints and limitations imposed by the Nao humanoid robots. We present the modelling challenges towards different objects in the world in terms of their dynamics, namely the static landmarks (e.g., goal posts, lines, corners), the passive moving ball, and the controlled moving robots, both teammates and adversaries. We discuss the approaches to model each such type of object depending on its motion model. Although our presentation is based on the specifics of the RoboCup SPL, the challenges and approaches we present are general to any multi-robot world modelling problem, as we state them in terms of classes of objects which should be part of general scenarios. 1.