Towards a versatile opportunity awareness algorithm for humanoid soccer robots using Time Petri nets

Abstract

To keep up with the advancements in robotics, techniques for computing and modeling of behaviors must be developed and researched. Using Petri nets (PNs) to model complex systems allows for analysis and increased comprehension of the system itself. PNs can be used to show the logic behind a system or to create a visual representation of the steps a system will take. Although this alone can be useful, there are some areas in which normal PNs fail; in these cases introducing time into the nets can open modeling possibilities. Time can be used to literally represent the duration of an event or can be used to implement probability into a system. In this paper, Time Petri nets (TPNs) are used to model soccer playing robots whose movements are based on the proposed Selective Kick Opportunity Awareness Response (SKOAR) algorithm, which serves to guide the robots to the safest path to the goal. After modeling, simulation and testing, it was shown that the proposed algorithm outperformed both the Soccer Playing Allies Referencing Tract and Coordinate Underlay System (SPARTaCUS) algorithm and the Rapidly-exploring Random Tree (RRT) algorithm in goal score rate by about 3% and 7%, respectively and in goal attempt success rate by about 5% and 17%, respectively.