Learning and stabilization of altruistic behaviors in multi-agent systems

Approximating an expectation related to cooperative robots

“…The model is the following (one can consult [1] for more details and [2][3][4][5] for related works in cooperative domains):…”

Probability distribution of the number of deceits in collective robotics

“…to cooperate or not) in a self-interested way. Furthermore, when all members of a collective behave in a cooperative way, reaching optimal performances frequently requires the presence of individuals who behave altruistically [11]. However, an important shortcoming arises.…”

“…Consequently, knowledge of the probability models of these variables is essential in determining the viability of altruistic cooperation. In artificial societies (either simulated or real robot societies) reciprocity has been successfully modeled to stabilize altruistic behavior [8], [11]. In [11], a group of self-interested robots optimize a gathering task, learning the type of cooperative behavior to show.…”

“…number of cooperative robots, probability of deceit recognition, etc.). As a contribution to overcoming this lack of knowledge, we develop a probabilistic model of the performance of a selfish robot in the experimental conditions described by Zamora et al [11].…”

“…Furthermore, under some circumstances, the number of robots is also modifiable. The fact that the mean value of this important variable X n,q depends on modifiable parameters allows the designer of such a system to control the egoistic benefit and, thus, the viability of cooperation (see [11] for a more detailed experimental setup description and [6] for a description of the robot). The relevance of this model can be summarized as follows:…”