This paper analyzes the perception of emotions and the attribution of human characteristics to the PLEA robot based on nonverbal communication. The aim of the study was to examine how individuals interpret the robot's emotional expressions, which emotions they share with it and which emotions they perceive as most convincingly communicated. The research was conducted through an experiment where participants nonverbally communicated with the PLEA robot using facial expressions, followed by a survey describing their perception of the PLEA robot. The data collected through the survey were organized using descriptive statistical tools. Additionally, general assumptions or hypotheses were formulated before the experiment and later tested through analytical processing and data visualization. PLEA is an interactive robot developed at the Faculty od Mechanical Engineering and Naval Architecture, designed for nonverbal communication with humans through facial expressions. The robot recognizes participants' facial expressions, categorizes them into appropriate emotional categories, and responds by displaying its own emotions. Due to this ability to interpret and mirror emotional expressions, PLEA is considered a reactive emotional mirror. The data collection process consisted of two stages. The first stage involved the interaction of participants with the PLEA robot, while the second phase of the experiment consisted of a questionnare, which the participants completed after the interaction, describing their experiences and impressions during the interaction. The collected data were analyzed using a combination of data processing methods and statistical analysis. Sentiment analysis of interview responses was conducted using the DistilBERT deep learning model, while quantitative data were processed using descriptive statistics. This approach provided a more detailed insight into the perception of the robot's emotions and the degree of anthropomorphization observed among participants.

Ovaj rad bavi se analizom percepcije emocija i pridavanja ljudskih osobina robotu PLEA temeljem neverbalne komunikacije. Cilj istraživanja bio je ispitati kako ljudi interpretiraju emocionalne izraze robota, koje emocije djele s njim te koje emocije smatraju najuvjerljivije komuniciranima. Istraživanje je provedeno kroz eksperiment u kojem su sudionici neverbalno komunicirali s PLEA robotom koristeći izraze lica, a potom su odgovarali na anketna pitanja koja opisuju njihov doživljaj PLEA robota. Podaci prikupljeni kroz anketu su organizirani koristeći alate opisne statistike. Također su prije provođenja eksperimenta postavljene opće pretpostavke odnosno hipoteze koje su zatim testirane kroz analitičku obradu i vizualizaciju podataka. PLEA je interaktivni robot, razvijen na Fakultetu strojarstva i brodogradnje, a dizajniran za neverbalnu komunikaciju s ljudima putem izraza lica. Robot prepoznaje izraze lica sudionika, svrstava ih u odgovarajuće kategorije emocija te na njih reagira prikazivanjem vlastitih emocija. Zbog ove sposobnosti interpretacije i uzvraćanja emocionalnih izraza, PLEA se smatra reaktivnim emocionalnim zrcalom. Proces prikupljanja podataka za analizu proveden je u dvije faze. Prva faza je uključivala interakciju ispitanika s PLEA robotom, dok je druga faza eksperimenta bio upitnik, kojeg su ispitanici popunjavali nakon interakcije, a u kojem su im postavljena pitanja koja opisuju njihova iskustva i dojmove tijekom interakcije. Prikupljeni podaci su analizirani kombiniranjem metoda obrade podataka i statističke analize. Za analizu sentimenta odgovora iz upitnika korišten je model dubokog učenja DistilBERT, dok su kvantitativni podaci obrađeni primjenom statističke analize. Ovaj pristup je omogućio detaljniji uvid u percepciju emocija robota i stupanj antropomorfizacije uočene kod ispitanika.