Research output: Contribution to journal › Article › peer-review
Adaptive control system for a mobile agent in a physical environment based on functional systems theory. / Putintsev, N. I.; Isupov, O. V.; Vityaev, E. E.
In: Russian Journal of Genetics: Applied Research, Vol. 5, No. 6, 01.11.2015, p. 601-608.Research output: Contribution to journal › Article › peer-review
}
TY - JOUR
T1 - Adaptive control system for a mobile agent in a physical environment based on functional systems theory
AU - Putintsev, N. I.
AU - Isupov, O. V.
AU - Vityaev, E. E.
PY - 2015/11/1
Y1 - 2015/11/1
N2 - We previously developed an adaptive mobile agent control system based on functional systems theory and semantic probabilistic inference (Vityaev, 2008; Demin and Vityaev, 2008). In the present work, we extended the potential of the system by introducing the ability to control a robot in physical environment. On the one hand, this ability shows that the system can operate in the environment that animals function in. On the other hand, it allows testing of the developed algorithm on actual physical environment. We identified two objectives. The first was to extend the capabilities of the system so that it could operate effectively in the physical environment; in particular, it was necessary to add support for continuous sensors and carry out a simulated experiment. The second was to extend the semantic probabilistic inference for the case of continuous sensors. The system was supplemented with abilities to use sensors with continuous real signals and to vary the duration of its actions when selecting a way to achieve the goal. The benefits of the semantic probabilistic inference were preserved. We constructed a robotic platform for experiments in the physical environment. The platform could carry several types of sensors and move according to commands received wirelessly. To show the ability of acting in the physical environment, the system was supposed to learn how to find bricks scattered around the room. The developed algorithm made it possible to solve this task and generate a set of rules for the effective detection of bricks.
AB - We previously developed an adaptive mobile agent control system based on functional systems theory and semantic probabilistic inference (Vityaev, 2008; Demin and Vityaev, 2008). In the present work, we extended the potential of the system by introducing the ability to control a robot in physical environment. On the one hand, this ability shows that the system can operate in the environment that animals function in. On the other hand, it allows testing of the developed algorithm on actual physical environment. We identified two objectives. The first was to extend the capabilities of the system so that it could operate effectively in the physical environment; in particular, it was necessary to add support for continuous sensors and carry out a simulated experiment. The second was to extend the semantic probabilistic inference for the case of continuous sensors. The system was supplemented with abilities to use sensors with continuous real signals and to vary the duration of its actions when selecting a way to achieve the goal. The benefits of the semantic probabilistic inference were preserved. We constructed a robotic platform for experiments in the physical environment. The platform could carry several types of sensors and move according to commands received wirelessly. To show the ability of acting in the physical environment, the system was supposed to learn how to find bricks scattered around the room. The developed algorithm made it possible to solve this task and generate a set of rules for the effective detection of bricks.
KW - adaptive control system
KW - functional systems theory
KW - machine learning
KW - mobile agent
UR - http://www.scopus.com/inward/record.url?scp=84949977354&partnerID=8YFLogxK
U2 - 10.1134/S2079059715060131
DO - 10.1134/S2079059715060131
M3 - Article
AN - SCOPUS:84949977354
VL - 5
SP - 601
EP - 608
JO - Russian Journal of Genetics: Applied Research
JF - Russian Journal of Genetics: Applied Research
SN - 2079-0597
IS - 6
ER -
ID: 25327849