Detail publikace

LEARNING BASED CONTROL SYSTEM OF FOUR-LEGGED ROBOT

BŘEZINA, T. HOUŠKA, P. SINGULE, V.

Originální název

LEARNING BASED CONTROL SYSTEM OF FOUR-LEGGED ROBOT

Český název

LEARNING BASED CONTROL SYSTEM OF FOUR-LEGGED ROBOT

Typ

článek ve sborníku

Jazyk

cs

Originální abstrakt

Possible discretization technique of the continuous state space of four-legged robot using simultaneous compositions of behaviors is described in this contribution. Compositions are generated by the instances of two basic controllers. The aim is to automatically develop the gait policy. Possible composition strategies are implemented through undeterministic state machine. In the machine design stage the number of both machine states and transitions could be essetially reduced using formal considerations only. Furthermore machine transitions are evaluated by selected alternative of Q-learning to incrementally improve the estimation of the future benefit from single transitions. This technique results in simultaneous activations of controller compositions instances with the highest estimation of the future benefit.

Český abstrakt

Possible discretization technique of the continuous state space of four-legged robot using simultaneous compositions of behaviors is described in this contribution. Compositions are generated by the instances of two basic controllers. The aim is to automatically develop the gait policy. Possible composition strategies are implemented through undeterministic state machine. In the machine design stage the number of both machine states and transitions could be essetially reduced using formal considerations only. Furthermore machine transitions are evaluated by selected alternative of Q-learning to incrementally improve the estimation of the future benefit from single transitions. This technique results in simultaneous activations of controller compositions instances with the highest estimation of the future benefit.

Klíčová slova

Control Composition, Q-learning, Walking of Robot

Rok RIV

2002

Vydáno

13.05.2002

Nakladatel

Institute of Mechanics of Solids, Faculty of Mechanical Engineering, Brno University of Technology

Místo

Brno

ISBN

80-214-2109-6

Kniha

Inženýrská mechanika 2002

Číslo edice

1

Strany od

9

Strany do

10

Strany počet

2

BibTex


@inproceedings{BUT9660,
  author="Tomáš {Březina} and Pavel {Houška} and Vladislav {Singule}",
  title="LEARNING BASED CONTROL SYSTEM OF FOUR-LEGGED ROBOT",
  annote="Possible discretization technique of the continuous state space of four-legged robot using simultaneous compositions of behaviors is described in this contribution. Compositions are generated by the instances of two basic controllers. The aim is to automatically develop the gait policy. Possible composition strategies are implemented through undeterministic state machine. In the machine design stage the number of both machine states and transitions could be essetially reduced using formal considerations only. Furthermore machine transitions are evaluated by selected alternative of Q-learning to incrementally improve the estimation of the future benefit from single transitions. This technique results in simultaneous activations of controller compositions instances with the highest estimation of the future benefit.",
  address="Institute of Mechanics of Solids, Faculty of Mechanical Engineering, Brno University of Technology",
  booktitle="Inženýrská mechanika 2002",
  chapter="9660",
  institution="Institute of Mechanics of Solids, Faculty of Mechanical Engineering, Brno University of Technology",
  year="2002",
  month="may",
  pages="9",
  publisher="Institute of Mechanics of Solids, Faculty of Mechanical Engineering, Brno University of Technology",
  type="conference paper"
}