Analysis of a Visual Imitation Algorithm on a Robot Swarm

Authors : Ferhat Demiray, Mehmet Dinçer Erbaş

Pages : 1789-1803

Doi:10.29130/dubited.1390036

View : 88 | Download : 96

Publication Date : 2024-10-23

Article Type : Research Paper

Abstract :In this research, we examined a visual imitation algorithm on a group of real robots and analyzed the source of copying errors that are made by the robots visually learning by using this algorithm. As the two possible sources of the copying errors, the actuators of the demonstrator robot and the sensors of the learner robot were specified. First, it is calculated the amount and frequency of errors due to the actuators and we showed that errors due to actuators of the demonstrator robot were minimal. Second, it is examined the errors due to the sensors by using two different trajectory similarity metric in an experiment scenario and we discussed the origin of this kind of imitation error. In this way, we were able to model a source of behavioral diversity in a robot collective, which is similar to the natural systems, which results from errors that emerge during imitation activity.
Keywords : Görsel imitasyon, Çok robotlu sistemler, Sürü Robotlar