Upravljanje razgovorom s Nao robotom korištenjem POMDP metode

master's thesis

• Title (english): Management of Dialog with NAO Robot using POMDP Method
• Author: Barbara Kužet
• Mentor: Zdenko Kovačić (mentor)
• Committee member: Zdenko Kovačić (predsjednik povjerenstva)
• Committee member: Stjepan Bogdan (član povjerenstva)
• Committee member: Matko Orsag (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2016-07-14, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Electrical Engineering

Abstract

Cilj ovoga rada upravljanje je dijalogom NAO robota i djeteta u svrhu dijagnostike autizma. Upravljanje razgovorom ostvareno je POMDP modelom koji omogućava modeliranje neodređenih procesa kakav je i razgovor te praćenje stanja djeteta koje je neosmotrivo. Na temelju djetetovih odgovora donosi se ocjena njegovog stanja, a moguća su tri stanja djeteta: dijete je tipičnog razvoja, dijete ima poremećaj visoko funkcionalnog autizma ili ima poremećaj niskofunkcionalnog autizma. Odgovori djeteta prvo se klasificiraju u jednu od 6 određenih skupina, a zatim se šalju kao opservacije POMDP modelu koji u odnosu na njih određuje akcije robota. POMDP model pokazuje dobro vladanje za točnost prepoznavanja odgovora djeteta od 80% i više. Simulacijama razgovora za nekoliko scenarija, pokazano je da se izgrađenim modelom uspješno vodi razgovor i da automatski generirana ocjena stanja djeteta daje zadovoljavajuće rezultate.

Abstract (english)

The aim of this thesis is managing the dialogue between a NAO robot and a child for the purpose of autism diagnosis. Conversation managing was achieved by POMDP model which facilitates modeling of uncertain processes, such as conversing and monitoring the condition of a child that is unobservable. Based on the child’s answers the evaluation of its condition is made, and there are three possible states which can be proclaimed: the child is of typical development, the child has a high functioning autism, or the child has a low functioning autism. Child’s answers are first classified into one of six possible groups and are then sent as observations to the POMDP model, which sets the actions of the robot based on the given answers. POMDP model behaves adequately when answer recognition accuracy is 80 per cent or higher. Conversation simulations of a few different scenarios show that the built model converses succesfully and that the automatically generated assessments of the child give satisfactory results.

• Keywords: NAO robot; autizam; POMDP; NLU; SDS
• Keywords (english): NAO robot; autism; POMDP; NLU; SDS
• Language: croatian
• URN:NBN: urn:nbn:hr:168:402392
• Study programme: Title: Electrical Engineering and Information Technology; Study programme type: university; Study level: graduate; Academic / professional title: magistar/magistra inženjer/inženjerka elektrotehnike i informacijske tehnologije (magistar/magistra inženjer/inženjerka elektrotehnike i informacijske tehnologije)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2018-02-06 10:03:36