| Abstrakt: |
In systems engineering, requirements derivation is an important step to achieve a successful product design. This process consists of obtaining technical specifications that meet the client's needs. Some applications, however, do not have a clear relationship between these two. Therefore, we contribute with a requirements derivation method by studying the trade-off between product performance and specification demand. We apply this method to the RoboCup Small Size League, a robot competition that promotes research in artificial intelligence and robot systems. In this work, we focus in the goal interception task. The development of a team for this competition requires a hardware project to build the robots and a software project to make game decisions. In this context, we derive requirements both for the robot's parameters (maximum velocity, maximum acceleration, robot's bandwidth, robot's phase margin) and for the perception system's parameters (information update frequency, information delay). During this study's development, we model the robots and game dynamics to determine the influence of the system's parameters in the task and evaluate the goalkeeper's performance by using Monte Carlo methods to estimate the goalkeeper's performance. Moreover, we observe that the performance estimation depends not only on the system's parameters but also on the Monte Carlo sampling model and the selected game strategy. Lastly, realistic ranges for the design parameters are obtained using the proposed methods. [ABSTRACT FROM AUTHOR] |
Full text from SpringerLink