Napredna pretraga

Pregled bibliografske jedinice broj: 990276

A Reinforcement Learning Based Algorithm for Robot Action Planning


Švaco, Marko; Jerbić, Bojan; Polančec, Mateo; Šuligoj, Filip
A Reinforcement Learning Based Algorithm for Robot Action Planning // Advances in Service and Industrial Robotics. RAAD 2018. Mechanisms and Machine Science, vol 67. / Aspragathos, N. ; Koustoumpardis, P ; , Moulianitis, V (ur.).
Cham: Springer, 2018. str. 493-503 doi:10.1007/978-3-030-00232-9_52 (predavanje, međunarodna recenzija, cjeloviti rad (in extenso), znanstveni)


Naslov
A Reinforcement Learning Based Algorithm for Robot Action Planning

Autori
Švaco, Marko ; Jerbić, Bojan ; Polančec, Mateo ; Šuligoj, Filip

Vrsta, podvrsta i kategorija rada
Radovi u zbornicima skupova, cjeloviti rad (in extenso), znanstveni

Izvornik
Advances in Service and Industrial Robotics. RAAD 2018. Mechanisms and Machine Science, vol 67. / Aspragathos, N. ; Koustoumpardis, P ; , Moulianitis, V - Cham : Springer, 2018, 493-503

ISBN
978-3-030-00231-2

Skup
RAAD 2018: Advances in Service and Industrial Robotics

Mjesto i datum
Patras, Grčka, 06.-08.06.2018.

Vrsta sudjelovanja
Predavanje

Vrsta recenzije
Međunarodna recenzija

Ključne riječi
Robotics ; Reinforcement learning ; Autonomous robot

Sažetak
The learning process that arises in response to the visual perception of the environment is the starting point for numerous research in the field of applied and cognitive robotics. In this research, we propose a reinforcement learning based action planning algorithm for the assembly of spatial structures with an autonomous robot in an unstructured environment. We have developed an algorithm based on temporal difference learning using linear base functions for the approximation of the state-value-function because of a large number of discrete states that the autonomous robot can encounter. The aim is to find the optimal sequence of actions that the agent (robot) needs to take in order to move objects in a 2D environment until they reach the predefined target state. The algorithm is divided into two parts. In the first part, the goal is to learn the parameters in order to properly approximate the Q function. In the second part of the algorithm, the obtained parameters are used to define the sequence of actions for a UR3 robot arm. We present a preliminary validation of the algorithm in an experimental laboratory scenario.

Izvorni jezik
Engleski

Znanstvena područja
Strojarstvo



POVEZANOST RADA


Projekt / tema
HRZZ-IP-2013-11-4192 - Novi koncept primijenjene kognitivne robotike u kliničkoj neuroznanosti (Bojan Jerbić, )

Ustanove
Fakultet strojarstva i brodogradnje, Zagreb

Časopis indeksira:


  • Scopus


Citati