Some Mechanisms for Managing Aggressive Behavior in Group Robotics
Possible methods of implementing aggression as one of the mechanisms in the formation of social behavior in groups of robots are discussed. Aggression is seen as a way to resolve conflicts over resources. The features of the aggressive behavior of eusocial insects (ants) are used as a basis. The aggressive component is integrated into the need-emotional architecture of the robot control system, which is presented as a hybrid neuro-production system. The proposed mechanism can be used as a base for implementing various models of social behavior in group robotics.
One of the actively developing approaches of group robotics systems creation is the use of social behavior models. Aggressive behavior is one of the underlying mechanisms forming social behavior. In this paper, the application of aggressive behavior concepts is considered by analogy with animal aggressive behavior that can be used for solving tasks of group robotics. As a role model, an ant – a true social insect – is proposed. It was shown that in aggressive behavior of ants, the numerical factor and imitative behavior play an important role. Agent’s aggressive behavior model depending on accumulated aggression and the number of other nearby agents is proposed. The results of computer experiments for territory defense tasks are presented. The results show that aggression is a stabilizing factor for an approximately equal number of agents in different groups. By an increase in group size, aggression becomes a way of capturing foreign territory.
Bayesian Belief Networks (BBN) provide a comprehensible framework for representing complex systems that allows including expert knowledge and statistical data simultaneously. We explored BBN models for estimating risky behavior rate and compared several network structures, both expert-based and data-based. To learn and evaluate models we used generated behavior data with 9393 observations. We applied both score-based and constraint-based structure learning algorithms. The score-based structures represented better quality scores according to BIC and log-likelihood, prediction quality was almost the same for databased models and lower but sufficient for expert-based models. Hence, in case of limited data we can reduce computations and apply expert-based structure for solving practical issues.
Some possible ways of implementing aggression as one of the mechanisms for the social behavior formation in robots groups are discussed in this work. Aggression is considered as a way to resolve conflicts over resources. The features of the aggressive behavior of eusocial insects (ants) are used as basic. A reactive model of behavior was proposed. The aggressive component is integrated into the demand-emotional architecture of the animat's control system, which is presented as a hybrid neuroproduction system. Also, the question of using an aggressive component at the phenomenological level of behavior management. Imitation modeling experiments were carried out on the example of realization of domination in a group. The issue of determining the basic mechanisms for feed areas distribution, which is part of the foraging task, is also considered. It is shown that such mechanisms are domination (as the result of aggressive actions) and the animats memory. The simulation results confirm that the addition of the "aggressiveness" parameter to the control system provides a variety of animats behavior taking into account the environment state. The proposed model of aggressive behavior does not depend on the task, which is solved, and allows to manage the group in natural form.
The key regulator in the control of aggressive behavior is dopamine receptors. Association of variants in these genes with aggression has been shown in modern populations. However, these studies have not been conducted in traditional cultures. The aim of our study was to investigate population features in distributions of allele and genotype frequencies of DRD2 rs1800497, DRD4 120 bp Ins, and DRD4 exon III polymorphisms and their associations with aggressive behavior in the traditional African populations of Hadza and Datoga, which display a contrast in their culturally permitted aggression. Overall, 820 healthy unrelated Hadza and Datoga individuals were studied. Self‐rated scores of aggression were collected using Buss and Perry's Aggression Questionnaire. Polymerase chain reaction‐Restriction fragment length polymorphism (PCR‐RFLP) was used to determine the genotype of each individual. We show that the Hadza and the Datoga differed significantly in allele and genotype frequencies of all studied loci. Our association analysis detected that only ethnicity and sex of individuals significantly influenced their aggression rank, but we failed to identify any associations of DRD2 rs1800497, DRD4 120 bp Ins, or DRD4 exon III polymorphisms with aggression. Thus, our data have no strong evidence to support the involvement of polymorphisms of DRD2 and DRD4 in controlling aggressive behavior.
This paper discusses a problem of distributed data processing in mobile robot’s group. The robot’s group is typified as a static swarm. Static swarm is a model, which is characterized by the absence of a control сenter and is represented by the network with fixed topology at some time interval which consists of locally interacting agents. The main features that distinguish the robot’s group general database from the classic distributed database are described. It is shown that the database does not require the global data dictionary storing information about the location of the database fragments. The data structure on each node is the same and can be described in the reference table. This table is loaded into robot’s memory when robot is initialized and, in fact, is the data dictionary. This database does not require distributed transactions, because data is written in the general database locally. The approach of logical queries organization in general database is offered. Definitions of imprecise and inconsistent data conformably to data which robots in the group are exchanging with are given. The approach of processing imprecise and inconsistent data which come to robot is proposed. This approach is based on elements of multisets theory, fuzzy sets theory and on evaluation of data reliability degree. The reliability degree is based on the experience of the previous data exchanges between robots. An important feature of the proposed method of imprecise and inconsistent data processing is that it is computationally simple and does not require much memory to store auxiliary information.