مقاله انگلیسی رایگان در مورد یادگیری آنلاین در تیم های چند رباتی – IEEE 2019

In this manuscript, we propose an approach that allows a team of robots to create new (emergent) behaviors at execution time. Basically, we improve the approach called N-Learning used for selfprogramming of robots in a team, by modifying and extending its functioning structure. The basic capability of behavior sharing is increased by the catching of emergent behaviors at run time. With this, all robots are able not only to share existing knowledge, here represented by blocks of codes containing desired behaviors but also to creating new behaviors as well. Experiments with real robots are presented in order to validate our approach. The experiments demonstrate that after the human-robot interaction with one robot using Program by Demonstration, this robot generates a new behavior at run time and teaches a second robot that performs the same learned behavior through this improved version of the N-learning system.

Introduction

Brooks [1] was the first researcher to propose the concept of behavior-based robotics (BBR). This paradigm can be understood as a framework that uses a set of behaviors used by a group of robots. In BBR, a behavior selector chooses the appropriate behavior according to the current situation. The advantage of our approach is that the proposed architecture is modular-based, solving each problem separately by applying one or more behaviors. A behavior can be external when interacting directly with the environment, or internal when resulting in changes in the internal structures of a robot [2]. With this definition, we can create behaviors focusing on cognitive tasks [3]. The first time that the transferring (learning and teaching) of pre-programmed behaviors was proposed was in the work of Costa et al. [4], through the approach called N-learning. In the N-learning approach, behaviors are blocks of code with information about the execution of a specific maneuver or action, which can be shared throughout the multirobot team at execution time. The main objective of the approach is to enable a group of robots to share knowledge through their interactions. The knowledge is represented here as one or more behaviors that enable the robot team to adapt to situations that are not previously taught in its initial programming.