The use of systems of autonomous agents cover numerous applications in a number of activities. Thus, the system of autonomous agents is composed of agents that are interconnected to each other so that the system acts in the envisaged way in the environment. Autonomous agents are objects that independently and without human intervention execute the task set with some possible modifications and report on the course of action. Examples of autonomous agents are stationary or mobile robots, sailboats, unmanned aerial vehicles, agents in a virtual environment, and other types of agents that perform a smaller number of simpler operations. Activities performed by autonomous agents involve activities that are considered dangerous to humans, exhausting or boring. Examples of the use of autonomous agents are transport of goods and people between multiple locations, contactless and irrational search of sites, monitoring of natural phenomena, underwater research, traffic control, implementation of communication networks in adverse environments, and so on. While operating, agents in such systems constantly measure certain parameters related to their states or to the state of their environment. A general accompanying characteristic of these measurements is that they are processed with statistical limitations in real time. The dynamics of the system of autonomous agents follows several essential problems. The first problem is the lack of ways to determine the ability of the autonomous agents system. It is not known which combinations of dynamic characteristics of autonomous agents can accomplish the required action and achieve the given goal. Another problem is the impossibility of determining the robustness of such a system of autonomous agents. The ability to comprehensively assess the robustness and ability of the system will help solve the above mentioned problems. The ability of the system is defined as the ratio of the required range and the implemented system characteristics. The system is considered capable if this ratio is large enough. Accordingly, from the ability assessment of the system based on agent parameters, it is possible to determine which combination of agents are needed to achieve the target. The system's robustness refers to the system and environment relationship and indicates systems whose performance does not change significantly for a particular change in the environment [1-3]. The general characteristics of research system of autonomous agents is their empirical character and incompleteness. This is why a relatively small number of examples of the use of autonomous agent systems in real situations. Agents are constantly measuring more than one size during work. Intermediate communication and measurements get additional sizes. Collected sizes need to be processed in real time with all statistical constraints. If the system's capacity is empirically determined, it is necessary to carry out multiple tests, with the combined cost of time and energy consumption on average with higher environmental pollution and consequently higher cost of research. Due to the disadvantages of the current approach to the problem there is no complete scientific assessment of the robustness and ability of the system. Numerical simulations are relatively little represented in the relevant literature, and the results are based on repeated attempts and errors. Some authors consider and define the application of agents at a general level with examples of application [4-10] while other authors concentrate on specific applications [11-15]. Dorigo et al., based their research on the development of Ant Colony Optimization algorithm [4-6]. The same researchers apply systems agents for modeling and optimizing different systems such as the Ant Colony Optimization Algorithm for Continuous Optimization from Algorithmic Components [7] and solving the problem of a salesman traveler. In further research, the system of agents is applied to model a structured robotic group [8, 9]. Furthermore, they introduced Swarmanoid as a new concept in the application of a group of robots. They have developed hardware and software for the integration of different robots on the platform, i.e. the ability to join the group. The obtained compatibility of different robots enabled them to explore different coordination mechanisms and strategies in the heterogeneous group [10]. In the field of robotics research, Matarić applies a system of agents for the control and management of mobile robots in cooperative load transportation [11, 12]. Cooperative load loading was also investigated within the Aerial Robotics Cooperative Assembly system project. In the literature the authors focus on the construction and management of a group of mobile agents. Cooperative activities of the group were investigated by Mathew, Smith and Waslander and analyzed the ability of the group to deliver load. They analyzed cooperation between vehicle (truck) and unmanned aircraft (quad) in urban environment [13]. The path optimization was defined by the solution of the generalized problem of a salesman traveler. The activity of a group of autonomous agents to search and map the area was explored by Ducho et al. [16]. Co-operative activity of the group was also investigated by authors Hui et al. [15] and validated by Monte Carlo-simulation. Including consideration of capacity assessment system various authors [14, 17] apply Monte Carlo simulations [18] to validate realized modeled measurement uncertainty of the measuring system. The application of Markov's decision-making system to collective decision-making and collaboration between unmanned aerial vehicles was explored by Capitan et al. [19]. Interagency communication allows for self-organization in a dynamic environment, resulting in more stable navigation. The system is robust for errors in the operation of a smaller number of agents and is able to autonomously find the optimal path [20]. The scope of application of the system of agents system illustrates the exploration of natural phenomena such as the monitoring of wildlife migration by radio signals. In his research, Korner et al. use the unmanned aviation system to track migration on inaccessible terrains [21]. The use of unmanned aerial vehicles under such defined conditions ensures better connectivity, easier reading of signal strength, easier communication, faster locating of animals and easier recording of measured dimensions. In order to explore and map the areas due to the flood, Balta and his associates cite the cooperative application of unmanned aerial vehicles [22]. Except for soil surface research, agitation systems are also used in underwater research, so Sun et al. configure unmanned underwater vehicles as a system of agents [23]. The research has defined unmanned underwater vehicles system for underwater monitoring and avoidance of real-time obstacles. In most literature, systems are modeled, or empirically tested, as a group of agents. Thus, the starting quantities are defined for the individual agents and their averaging yields the size of the system. A qualitatively different approach is to use the size defined for the system, e.g. an interdisciplinary approach to socio-thermodynamic potentials [24, 25]. Analysis of the research area suggests that the determination of the ability of autonomous agent systems can be determined from the known dynamic characteristics of the agents and the characteristics of the environment in which the system operates. Taking into account the statistical measures associated with the characteristics of autonomous agents and the existing combination of set rules of communication as the input size, the question is how to in silico determine the ability of the system for a particular environment? Determining the system's ability to start up by setting the system model, which continues the numerical simulations of the set models. The ability to comprehensively assess the robustness and system abilities would remove the aforementioned drawbacks of the empirical approach and allow the operation of a larger number of systems in qualitatively different environments.