Task selection strategies of self-interested robots in skill games
Bao Fu Fang, Hao Wang, Ming Lan Fu · International Journal of Wireless and Mobile Computing · 2018
This paper focuses on task allocation for multiple self-interested robots coalition skill games. When the self-interested robots select tasks, it is difficult for them to maximise both their individual revenues and the system revenue simultaneously. But through the reasonable distribution of the tasks' utilities, the two matrices can be kept consistent to some extent. Based on this idea, an algorithm is proposed to allocate tasks to self-interested robots which can be in conservative state or radical state. The conservative task selection strategies ensure that the game will converge to Nash equilibrium. The radical task selection strategies help the algorithm jump out the Nash equilibrium. Tasks' utilities are distributed according to the robots' powers in the game. Meanwhile, the algorithm endows each robot with a numerical value, called patience, to allow the robots to transform between these two states. Finally, the simulation results verified the effectiveness of the algorithm.