Randomized controls for linear plants and confidence regions for parameters under external arbitrary noise

Combined procedure with randomized controls for the parameters' confidence region of linear plant under external arbitrary noise

“…It includes randomized items in the input channel only one time per some interval when the control strategy of [1] has randomized part of inputs on each iteration. This fact allows to prove in [12] more weak assumptions about independence of external noise and randomized part of control inputs.…”

“…Then justified recurrent algorithms such as stochastic approximation algorithms can be applied for estimating unknown values of the parameters. In this paper we combine the ideas of former asymptotic results from [2], [5] and new procedure from [11], [12] which gives rigorously guaranteed non-asymptotic confidence regions for unknown parameters of a linear dynamical control plant which is disturbed by arbitrary noise. The control strategy considered in [11], [12] is different from the control strategy of [1].…”

“…In this paper we combine the ideas of former asymptotic results from [2], [5] and new procedure from [11], [12] which gives rigorously guaranteed non-asymptotic confidence regions for unknown parameters of a linear dynamical control plant which is disturbed by arbitrary noise. The control strategy considered in [11], [12] is different from the control strategy of [1]. It includes randomized items in the input channel only one time per some interval when the control strategy of [1] has randomized part of inputs on each iteration.…”

“…Stochastic approximation algorithm is considered in Section VI. Next Section VII summarizes the result of [12] about properties of confidence regions in the case of a finite number of observations. The new algorithm and the main theoretical result of this paper are in Section VIII.…”

Task Allocation Algorithm for the Cooperating Group of Light Autonomous Unmanned Aerial Vehicles

“…During the flight, each UAV has its own number of tasks in a queue, and new tasks could be added to the queue over time. Moreover, productivities changes over time due to environmental influence such as, for example, the wind, which could be compensated by applying filtering techniques Amelin and Granichin (2012); Amelin (2012).…”