Skoči na glavni sadržaj

Izvorni znanstveni članak

https://doi.org/10.20532/cit.2017.1003573

Unibot, a Universal Agent Architecture for Robots

Goran Zaharija ; Faculty of Science, University of Split, Croatia
Saša Mladenović ; Faculty of Science, University of Split, Croatia
Lada Maleš ; Faculty of Humanities and Social Sciences, University of Split, Croatia


Puni tekst: engleski pdf 802 Kb

str. 31-45

preuzimanja: 571

citiraj


Sažetak

Today there are numerous robots in different applications domains despite the fact that they still have limitations in perception, actuation and decision process. Consequently, robots usually have limited autonomy, they are domain specific or have difficulty to adapt on new environments. Learning is the property that makes an agent intelligent and the crucial property that a robot should have to proliferate into the human society. Embedding the learning ability into the robot may simplify the development of the robot control mechanism. The motivation for this research is to develop the agent architecture of the universal robot – Unibot. In our approach the agent is the robot i.e. Unibot that acts in the physical world and is capable of learning. The Unibot conducts several simultaneous simulations of a problem of interest like path-finding. The novelty in our approach is the Multi-Agent Decision Support System which is developed and integrated into the Unibot agent architecture in order to execute simultaneous simulations. Furthermore, the Unibot calculates and evaluates between multiple solutions, decides which action should be performed and performs the action. The prototype of the Unibot agent architecture is described and evaluated in the experiment supported by the Lego Mindstorms robot and the NetLogo.

Ključne riječi

artificial intelligence, cognitive robotics, intelligent agents

Hrčak ID:

179238

URI

https://hrcak.srce.hr/179238

Posjeta: 964 *