Wyniki wyszukiwania - BazTech

1

The comparison of efficacy the gestures recognition algorithms based on RGB and RGB-D cameras for intuitive smart home control

Ambroziak A., Rodak M., Chojecki A., Borkowski P.

Zeszyty Naukowe Wydziału Elektrotechniki i Automatyki Politechniki Gdańskiej

|

2018

|

Nr 61

9--14

EN

This paper presents a comparison of the authors’ presence detection and gesture interpretation algorithm for a RGB camera with a commercial algorithm for a RGB-D camera for smart home control. The author’s presence detection algorithm is based on MOG2 algorithm employed for background learning and mathematical conversions for identification of position of an arm. The latter algorithm employs a depth camera to take pictures at which a human skeleton is overlaid and thus position of an arm in three-dimensional space is identified. The author’s algorithm enables to achieve better efficacy of gesture recognition (statistically of around 20%) employing less hardware resources at the same time. This makes it perfect for smart home automation control applications especially for people with disabilities.

PL

Artykuł prezentuje porównanie autorskiego algorytmu wykrywania gestu dla kamery RGB z algorytmem komercyjnym dla kamery RGB–D, na potrzeby sterowania inteligentnym budynkiem. Autorski algorytm oparty jest na algorytmie MOG2, wykorzystywanym do uczenia tła oraz w matematycznych przekształceniach w celu wykrycia pozycji ramienia. Na tej podstawie określany jest wektor wyznaczane przez rękę wskazującą na dany obiekt. Wskazywany element jest porównywany z bazą obiektów, następnie wykonywana jest związana z nim interakcja. Porównywany był on z komercyjnym algorytmem wykorzystującym kamerę głębi, która nakładała szkielet człowieka na obraz i pozycjonowała ramię w przestrzeni trójwymiarowej. Opracowany algorytm pozwala na osiągnięcie wymiernie lepszych wyników w skuteczności rozpoznawania (o ok. 20% w ujęciu statystycznym) w stosunku do komercyjnego algorytmu, przy mniejszym wykorzystaniu zasobów sprzętowych. Pozwala to na zastosowanie algorytmu w istniejących oraz nowo powstałych budynkach mieszkalnych, wykorzystując do jego implementacji mikrokomputery.

2

Impulse identification and discrete P/PD control of electro-hydraulic servodrive

Możaryn Jakub, Winnicki Arkadiusz, Suski Damian

Journal of Automation Mobile Robotics and Intelligent Systems

|

2018

|

Vol. 12, No. 4

51--57

EN

This article presents the systematic design methodology of a discrete Proportional and Proportional-Derivative (PD) controller for the electro-hydraulic servodrive position control. The controller is based on the identified linear model of the system, with P/PD parameters adjusted with the help of different methods given in the literature. There are compared experimental results of the proposed control system with different controller parameters.

3

An iterative identification based on the Youla-Kucera parameterisation without model reduction

Kozka S., Mikles J., Fikar M.

Archives of Control Sciences

|

2003

|

Vol. 13, no. 1

19-31

EN

The aim of this paper is to present an iterative identification of the plant in the presence of feedback using input-output data, based on the Youla-Kucera parameterisation. When a reduced complexity model is identified then the controller is designed. here the identified model is just a vehicle for the computation of a controller. The proposed iterative algorithm contains suitably selected filters and ensures that the model reduction of the identified model is not necessary as in the standard approach. This iterative approach simplifies the identification task of the Youla-Kucera parameter as only its numerator has to be estimated. an experiment with a continuous stirred tank reactor (CSTR), with represents a non-linear single input -single output (SISO)system, illustrate this feature. LQ controller design is used for stabilisation.

4

Some Issues in the Design of Predictive Controllers

Magni L., De Nicolao G., Scattolini R.

International Journal of Applied Mathematics and Computer Science

|

1999

|

Vol. 9, no 1

9-24

EN

In the paper, we discuss how to design a predictive controller capable of addressing a number of important issues ranging from nominal stability to the model identification/controller design interplay. Nominal stability is ensured by resorting to Constrained Receding Horizon Predictive Control. As for robust stability, the connections between the frequency weighting P-polynomial in the cost function and the achievable robustness against multiplicative uncertainty are investigated. Then, a two-step design procedure is proposed in order to enhance the closed-loop robustness and obtain nominal performances. A correlation technique is also proposed as a tool to estimate uncertainty bounds to be used in controller design. Finally, the control and identification procedures are put together to form an iterative identification/control design methodology. A simulation example is reported to illustrate the approach.