The drone-following models in smart cities: History Edit

Drones are estimated to play a critical role in the smart city with a variety of use cases: medical, transportation and agriculture. The applications of drones in the smart city will involve multiple drone platforms that operate simultaneously to run missions. The Federal Aviation Administration (FAA) predicted that 30,000 drones could be flying in U.S. skies in less than 20 years. Therefore, a safe and secure environment for drones' operational quality and stability is necessary.

The investigation of the drone traffic safety and the development of the intelligent transportation system needs drone-following models, which describes the one-by-one following process of drones in the traffic flow. There are two types of drone-following models that are proposed and discussed in this paper. The first models based on the principle that keeps a safe distance according to relative velocity, which based on the determining of the drone acceleration depending on the differences in speeds and gaps between the given drone and its leading drone. Another model is the Markov drone-following model which is an improved model based on the approximation of the stochastic diffusion process of speed decision. The simulation results show that the changes in velocities of the following drones are nearly the same as the leading one.

  • drones
  • following-drone models
  • drone models
  • traffic flow

A basis model based on determining the drone acceleration depending on the differences in velocities and distances between the given drone and its leading drone. The performance of the drones and controllers are defined by coefficient identified from the traffic flow data. In a simplified case, when the so-called optimal velocity models are used, the controllers maintain safe velocities according to the related position of drones. While, the so-called collision avoidance models describe situations, in which the controllers keep safe distances according to relative speeds.

The models describing the one-by-one following the process of drones in the traffic flow are called as drone-following models. They are one kind of microscopic simulation models applied in traffic system modeling and control. The drone-following models are expected to play essential roles in the development of the intelligent transport systems and collision-warning and collision-avoidance. The drone-following models must describe the process demonstrated in Fig. 1. The realized changes in velocity and position depend on the drone characteristics, speed and distance between the drones, controller parameters and weather conditions. The drone velocity and position depend on the traffic situation namely on the distance to the drone ahead and its velocity.

The drone-following models are developed for (I) producing the realistic speed profiles, (II) ability for generating the real traffic streams, (III) stable realization of the traffic situations, (IV) simulation of traffic situations realized by different combination of drones and controller parameters, and (V) ability to applied them in traffic control systems.

A.    The drone-following models based on a safe distance

The first drone-following models based on the principle that keeps a safe distance according to relative velocity (SD models). This approach led to the linear models assuming that the controller of the following drone controls the accelerator to keep relative speed to the leading drone.

 

It is desirable for the interval between successive recalculations of acceleration, speed, and location to be a fraction of reaction time, which necessitates the storage of a considerable quantity of historical data if the model is to be used in a simulation program. Moreover, the parameter has no apparent connection with identifiable characteristics of controller or drone.

The models are derived by setting limits on the performance of controller and drone and using these limits to calculate a safe speed concerning the former drone. It is assumed that the velocity of the following drone is controlled by a controller to ensure that the drone can hold safety when the leading drone held suddenly. The characteristics of the models are i) The following drone will accelerate if the leading drone accelerates, and ii) The following drone will decelerate and keep long distance if the leading drone decelerates and the distance between drones is short.

The model executes the inference system in small time increments. At each time increment, the action of leading drone can be changed; for instance, in one-time increment, it accelerates at a given rate; at the next time interval, it accelerates at another rate. The speed and position of the following drone relative to the leading drone are then updated after each time increment.

B.    The drone-following models based on Markov chain process

The SD models can be used widely for the traffic situation simulation. However, they have two disadvantages: i) the constants applied in the models very depend on the real traffic situations and drones and quality of controllers, ii) they are not taking into account the advanced controllers.

An improved model based on the approximation of the stochastic diffusion process of speed decision. The inputs of controllers receive the required changes in velocity from differences in speed and deviations in relative distance between the drones.

The Markov type of drone-following model improved by using the:

  • exponents, cv and cx for relative velocities and distances,
  • time reaction, T, depending on the actual relative distances, ΔXn[k],
  • controllers’ parameters depending on the actual reaction time,
  • estimated nominal values, ΔXpdn for the relative distances.

 The improved model can be used for simulation and analysis of the chaotic processes appearing in the drone-following processes. The chaotic processes depend on the parameters applied in the drone-following models.

With accordance to our investigation the chaos in drone-following processes mainly caused by the time delay (controllers’ time delay).