The authors present a problem of the performance of Unmanned Aerial Vehicles (UAV)'s flights (group or single flight) for the decision of different target tasks in the city using information air navigation technology and methods of mathematical modeling in Artificial Intelligence (graph theory, Expert Judgment Method, methods of decision making in risk and fuzzy-logic, dynamic programming, etc.). The configuration and optimization of group flight routes for UAVs depend on the type of "target task". The algorithm of estimation performance of UAVs flights in the smart-town, an illustrative example of the optimization of UAVs flights is presented in the article.
Remotely piloted aircraft systems (RPAS) are a new component of the aviation system. They are based on cutting-edge developments in aerospace technologies, which may open new applications; improve to the safety and efficiency of aviation [1; 2].
Unmanned Aerial Vehicles (UAV)'s have several advantages, namely low
operating cost, simplicity, availability, UAVs may be used in cases where the usage of
manned aircraft is impractical, expensive or dangerous [3; 4]. Nowadays using of
UAVs is effective for decision lot problems such as in monitoring forest fires; search
and rescue operations; for relay communications in those places - where the antenna
coverage cannot be set because of difficult terrain; in logistic as the safest, cheap and
fast method of movement of goods; for aerial photography; for controlling traffic; for
first aid to people under various extreme conditions, etc. [3; 4; 5]. Many of these tasks
decision for an urban locality and wherein effectively use single and group flight of
UAVs [6; 7]. The Forum "Urban Air Mobility" in November 2018 at Amsterdam
discussed the future of drones in cities. Looked at from the perspective of cities and
citizens, urban air mobility and the idea of Mobility as a Service (MaaS) provide a
fascinating view of a possible future where a daily commute could seamlessly include
a bicycle, train, and drone service all as part of an integrated public transportation
system [
When planning UAV flights, it is important to comply with regulatory air navigation requirements and effective methods for flight operations [1; 2; 9]. The documents of ICAO are including the requirements and UAV management rules such as UAV certification and operator certification; UAV registration; rules for UAV operations; communication with the UAV; training of personnel for the operation of the UAV; emergency situations with UAV and flight safety; legal issues to ensure the possibility of performing safe, coordinated and effectively integrated flights UAVs [1; 2].
The purposes of the work are: building an Expert system (ES) as Artificial Intelligence (AI) for estimation of the performance of UAVs flights (group and single) for the decision of different target tasks in an urban locality; definition safe and minimal cost ways UAVs movement in town. 2
Flexible redistribution of autonomous Unmanned Aircraft routes in multilevel airspace 2.1
Expert systems for estimation performance of UAVs flights in smart-town The concept of “smart city” is characterized by using the new achievements for the effective organization of life in a town. This is using AI as UAVs and Expert systems; Internet technologies in order to monitor the state of urban infrastructure facilities, their control, and based on the data obtained because of monitoring, optimal allocation of resources and ensuring the safety of citizens. Such objects include bridges and tunnels, roads and railways, communication systems, water supply, and drainage systems, power supply systems, and various large industrial facilities, airports, rail railway stations, seaports, etc. [4; 9].
The effectiveness of presenting using UAVs for a modern town as a “smart city”
has some problems: the presence of buildings, roads, construction, recreation areas,
and natural areas, etc.; availability of specific flight orders - target use of drones [
Using graph theory can determine the effectiveness of different structures
(topologies) in UAV’s group formation. To control a group of drones from RPAS suggested
choosing and using a Central Drone Repeater (CDR) to connect to the operator on the
ground and control the other of the UAVs using the method of server selection in
local computer networks [6; 10]. For planning and flight control UAV developed a
Distributed Decision Support System (DDSS), which represents a complex system
with complex interactions geographically distributed local Remote piloted aircraft
(RPA). During the flight UAVs may be controlled by remote piloting station (RPS).
At any given time ti k-UAV must piloted by only one j-th RPS, if necessary, at time
ti+1 to be transmitted to the control (j + 1)-th RPS (fig. 1). This transfer flight control
of the j-th RPS to (j + 1)-th RPS to be safe and effective, which is provided through
the local operators UAV. To coordinate interaction and exchange of information
between remoted pilots developed a database of local RPS NoSQL [
The authors have developed computer programs for DDSS of the unmanned
aircraft pilot, “Remote Expert Air Traffic Management System “Decision making (DM)
in a common environment FF-ICE (Flight & Flow Information for a Collaborative
Environment (FF-ICE)” presented decentralized-distributed UAVs control system
using blockchain technology for connection between RPS and RPA (Fig.1).
Blockchain technology is ideal as a new infrastructure to secure, share, and verify learning
achievements and Collaborative Decision Making (CDM) too. For today, the key to
ensuring the safety of flights is the problem of the organization of CDM by all the
operational partners based on general information on the flight process and ground
handling of the manned and unmanned aircraft [
For the management of the UAV, a system for managing single or a group of the UAV’s is proposed, depending on the purpose of the UAV (“target task”). Taking into account the limited and dependence of the use of the UAV group on its intended purpose were analysed the network topology indicators for the implementation of the group flight. The Algorithm of building an Expert system (ES) for estimation of the performance of UAVs flights (group and single) for the decision of different target tasks in an urban locality:
1. The estimation effectiveness performance the target task of using the next systems: the group of separate UAVs with controls from separate operators; the UAVs group with control from CDR-UAV; single UAV with control single operator. If there is the UAV group with control from CDR need: a. Decomposition of the complex system on subsystems “network topologies - the target tasks”, description of the characteristics of subsystems, and estimation of effectiveness of network topologies for performance the specific target task. b. The effectiveness of network topologies for performance the target task and definition of criteria estimation (definition the corresponding weight coefficients of the efficiency of the topology). c. Estimation of network topologies of the UAV group for the specific target task using Expert Judgment Method (EJM) (definition of system preferences and coordination of experts’ opinions too).
2. Estimation of urban locality using GRID analyses of sector UAV flight, fuzzylogic or EJM for estimation of risk/safety of UAV flight.
3. Aggregation of subsystems to the new system (additive or multiplicative aggregation depend on the type of "target task").
4. Graphical presentation of results for Expert System (group UAV, single UAV or group of single UAVs), for example, estimation of effectiveness of network topologies for performance the target task “monitoring" by UAVs group” (Fig.2).
To evaluation, the safety of UAVs flights in town, need to obtain quantitative
values of risks of flights in different segments of the territory of the town using methods
for evaluating risk/safety (EJM or Fuzzy logic) [
The air navigation rules for classification obstructions in town such as “Restricted” and “Dangerous” areas, but they have nothing in common with ICAO’s official definitions, this is an estimation of risks movement ways of UAVs in smart-city. The "Restricted areas" in our case are such areas, where the risk of harming people is high, the “Dangerous areas” - the risk of harming people is very high. Initial data for estimation risk:
a). Buildings. These are objects where people live and work (offices, factories, markets) and public places. Potential risk after these area penetrations: for UAVs very high; for people - moderate to high.
b). Columns and wired communication. These objects are columns with its wires, masts, pipes antennas, which may endanger life and health of people nearby in case of breakdown. Potential risk after area penetration: for UAVs - moderate; for people low to moderate.
c). Trees and natural obstructions: These objects are trees, hills, mountains etc. Potential risk after area penetration: for UAVs - high to very high; for people - very low.
d). Dangerous areas are classified on the basis of an application to the object of “Restricted area”. “Dangerous areas” themselves are not hazardous, but permanent residence increases the risk directly proportional to the residence time. Potential risk at the moment of penetration: for UAVs - very low; for people - very low.
e). The potential risk when UAVs is staying in any period of time is a very complex task and depends on many factors, such as time, enclosing object, the previous trajectory of flight, maneuverability of UAVs, aerodynamic aspects, environmental conditions, etc.
f). Track area. It is a part of the planned flight path after UAV flight in which 99.99% UAV is or will be located according to “Flight plan” data: for UAVs - high to very high; for people - high to very high.
g). Track conflict area: It is unplanned part of space around “Track area”: for UAVs - high to very high; for people - high to very high.
The results of values of risk/safety estimation of UAV flights in the city presented in Table 1. For example, risk of UAV flight in a restricted area equal to ten conventional units (multiplication the hazard/safety flight weight by the expected damage).
Fuzzy logic methods have been applied to assess risk levels and is based on the
logical rules "IF (condition) - TO (conclusion)" [
R (Pj ) (Lij ) The qualitative risk level indicator includes next characteristics of risk, namely: 1. “Very low risk” corresponds to the flight of UAV. 2. “Low risk” corresponds to restricted areas such as columns and wired communication; 3. “Average risk” corresponds to restricted areas such as a building; 4. “High risk” corresponds to dangerous areas; 5. “Very high risk” corresponds to the tracks area by busy of UAV.
The degree of belonging of a certain value determined as the ratio of the number of responses in which the value of the linguistic variable occurs in a certain interval, to the maximum value of this number in all intervals.
Experts were interviewed by the Delphi method in two rounds. There are 35 experts attend the survey. The results of the survey are listed in Table 4. Units of intervals – 1 for 0 - 0,1; 2 for 0,1- 0,2, 3 for 0,2- 0,3,etc. cij
The membership functions for estimation of risk were obtained based on experimental data. Assume that the minimum risk level is zero units and the maximum is 100 units respectively. The fuzzy-logic functions of estimation in risk moving UAVs in flight, track conflict area, track area, restricted area, and dangerous area in Fig.3 (after the first round of the poll).
From the resulting diagrams, determined the quantitative indicators that correspond to the values of the linguistic variable "risk level"(after the second round of the poll): “Very low risk” corresponds to the quantitative significance of the level of risk in 10. “Low risk” corresponds to the quantitative significance of the level of risk in 35; “Average risk” corresponds to the quantitative significance of the level of risk in 60; “High risk” corresponds to the quantitative significance of the level of risk in 80; “Very high risk” corresponds to the quantitative significance of the level of risk in 100. 2.2
Definition minimal cost and safety of UAVs movement ways in town The mathematical methods such as the Dynamic Programming (DP), EJM, and fuzzy logic for estimation risks and minimal cost of ways of moving. For a definition, minimal cost and safety of UAVs movement ways in smart-city of town may use mathematical methods and modern air navigation rules. Estimation of an area in a fragment of the territory in fig.4a. Algorithm of definition minimal cost and safety of UAVs movement ways in town next: 1) Grid-analysis - cells are superimposing on a fragment of terrain (Fig.4b). 2) Risk assessment of Grid cells depending on the type of area (“Restricted” or “Dangerous”).
3) Finding the minimum cost path W1 for a UAV1 using the DP method for planning a flight in a level L1:
Wi (yi ) yi1(RA; BA; TA; TCA; FA) min yi (RA; BA; TA; TCA; FA) a b Fig. 4 Fragment of the territory for estimation minimal cost and safety of UAVs movement Assessing the path W1 (level L1) of the UAV1 as “Dangerous”; 4) Finding the minimum cost path W2 for a UAV2 using the DP method for planning a flight in a level L1, if necessary, the transition to the level L2, etc.
For example, estimation and finding the minimum cost path W1 for a UAV1 on Fig.5, and the minimum cost path W1 for a UAV1 on Fig.6 (W1=39).
The transfer of a UAV flight from level L1 to level L2 is shown in Figure 7 when loading the first level.
Flow optimization and flexible redistribution of autonomous UAV routes in multilevel airspace is performed in accordance with air navigation rules. The documents of ICAO include main recommendations for using UAVs, i.e. the operation of the UAV should minimize the threat of harm to life or health of people, damage of property, danger to other aircraft [1; 2; 11]. 3
Further research should be directed to the solution of practical problems of actions UAV’s operator in case of emergencies, software creation. The organization of CDM by all aviation operators using collaborative DM models (CDMM) based on general information on the flight process and ground handling of the UAVs. Models of flight emergencies (FE) development and of DM in Risk and uncertainty by UAV’s in FE will allow predicting the operator’s actions with the aid of the Informational-analytic and Diagnostics complex for research UAV operator’s behavior in extreme situation.
For example, the synthesis of models for DM in an emergency if is solving logistic problem UAV flight in bad weather condition (emergency - "loss connection"). (in Figure 7). In the process of analysis and synthesis of DM models of AI in emergency tend to simplify models (stochastic, the neural network, fuzzy, the Markov network, GERT-models, reflexion models to deterministic models).
In order to simulate DM under conditions of an emergency, next steps: an analysis of an emergency; intelligent data processing; analysis and identification of the situation using stochastic models; decomposition of the situation as a complex situation into subclasses and the formation of adapted deterministic models of AI actions are made. The models for decision and predicting of EF using CDMM – technology presented in Table 2.
In cases of big and difficult data methods can be integrated into traditional and
next-generation hybrid DM systems by processing unsupervised situation data in the
deep landscape models, potentially at high data rates and in near real time, producing
a structured representation of input data with clusters that correspond to common
situation types [
Fuzzy logic to determine quantitative estimates of
potential loss - functions of estimation risk R /
outcomes U for next models of DM in Risk and
Uncertainty-{gr}
DM in Risk. Stochastic models types’ tree, GERT’s
network (Graphical Evaluation and Review
Technique) for DM and FE developing. The optimal
solution is found by the criterion of an expected
value with the principle of risk - Adopt
DM in certainty using Network Planning method
and DM in Risk for each branch. Determined
models for an operators / AI with deterministic
procedure - ti; ;Тcr;Тmid;Тmin;Тmax
Optimal decision for action in EF (operator / AI
model). The authors have developed a computer
program for finding optimal solutions [
It was presented a problem of the performance of UAV’s flight plans for group flights or single flights for the decision of different target tasks in the city (monitoring, data acquisition, transportations, urban survey, etc.) using information technology, graph theory, and mathematical methods. The configuration and optimization of group flight routes for UAVs depend on the "target task" and results of estimation (cost/safety) territory for UAVs flights. The algorithms of building an ES for estimation of the performance of UAVs flights (group and single) in an urban locality and definition ways of minimal cost/safety of UAVs movement in town were presented. Further research should be directed to the solution of practical problems of actions UAV’s operator / AI models in case of emergencies and software creation according to the target task. Next planned to use new methods for DM (Big Data, Blockchain technology, AI models, next-generation hybrid DM systems; Data mining, etc.).