Search

Can humans trust humanoid robots to perform social tasks? Trust is a multifactorial concept with several determinants that influence the interaction between humans and robots. These determinants include factors concerning human, robot, social, and context. To measure human trust in humanoid robots, interactive dialogues were used with subjective measures of general trust embedded in the dialogues. General trust was made up of three components: ability, benevolence, and integrity, each characterized by five attributes. These measures were mapped to objective measures of physiological trust comprising facial expressions, voiced speech, vision-based heart rate, and postural gestures. The purpose was to predict trust from a neural network of psychophysiological criteria. To validate the reliability of trust measures, a series of three experiments was conducted at three levels of interaction: human-human, human-robot and human-robot-human. While the methodology was used to measure human trust in robots, it has potential for measuring robot trust in humans.

This paper describes a methodology for exploring trust using psychological (subjective) and physiological (objective) correlates to trust. The aim was to explore trust using natural dialogs of real-world scenarios that embed fifteen subjective measures. The goal was to apply the method in modeling human-robot-human interaction, involving three types of androids and to predict trust. Two forms of dialogs were employed: a guided script and a predetermined dialog representing three social scenarios. Objective features included facial expressions, voice and heart rate. Subjective trust measures comprised ability, benevolence and integrity. A repeated measures experimental design was employed. Forty-two subjects participated in the study. The data was analyzed using exploratory factor analysis and correlation. Multiple neuro-fuzzy models were trained using the data set and combined as an ensemble using evolutionary algorithms. The final ensemble estimated trust with 67% accuracy. The implications of the findings and limitations of the method are discussed.

We analyzed customer expertise in purchasing a car that satisfies customer needs for affective design. The approach was based on a hierarchical patterns analysis. There is a hierarchy of requirements starting with simple marketing requirements and ending with final design details. We assume that there is a rationale for customer’s evaluation of cars, and that evaluations are connected to design details. The purpose of using a hierarchical pattern was to model how car buyers evaluate car appearance. In addition, a hierarchical pattern is helpful to establish a dialog between car buyers and designers and create a frame work, where customer requirements can be satisfied. The study examined five main roles of product appearance, namely: Aesthetics, Symbolic Meaning, Functionality, Ergonomics and Categorization. The functional and ergonomics role are described as semantic roles as they address user’s evaluation of the apparent utility and perceived quality of a design. Twenty-six sets of evaluations were obtained from interviews with 13 subjects. To arrive at hierarchical patterns, the data was analyzed in three steps: Step 1 involved classifying responses into three groups: “I like”, “I dislike” and “I want to modify”. Step 2 transformed the results into hierarchical patterns using questions such as “Why do you like this car?” Step 3 analyzed and generalized the hierarchical patterns into a few simple patterns. The analyses yielded seven different hierarchical patterns, which generally fit all the evaluations of cars. The results are discussed in the context of the product appearance roles.

The measurement of human trust in humanoid robots in human-robot interaction requires novel approaches that can predict trust effectively. We present a method that mapped subjective measures (i.e. general trust, psychological) to objective measures (i.e. physiological) to predict trust. We designed interactive dialogs that represent real world service scenarios of Business, Disaster, and Healthcare. The dialogs embedded fifteen trust attributes of Ability, Benevolence and Integrity (ABI) in the communication dialogs. The ABI measures were mapped to physiological measures of facial expressions, voiced speech and camera-based heart rate. Forty-eight subjects comprising 24 males and 24 females aged between 18 to 36 years participated in the experiment. Half of the subjects were Malays and half were Chinese. Three humanoid robots represented full bodied, partial bodied and virtual agents. The experimental design was a within-subjects design. Each subject was tested on all robots in all scenarios. Subjects scored trust on an online scale that ranged from 0 to 7 points. The subjective data was analyzed using Univariate and Oneway MANOVA. The results found the humanoids to be trustworthy in different service tasks. The attributes of ‘Integrity’ and ‘Ability’ trust components are important in Business and Disaster scenarios. The estimation of trust was about 83% accurate when using this creative approach. In conclusion, humanoid robots can interact with humans using dialogs that are representative of real world communication.

Trust is a quality representing a person's willingness to rely on a trusted party to cooperate. Quantification of trust is highly subjective since it relies on self-reports. In this study, we analyzed trust between human subjects in terms of objective measures of facial expressions, voice and heart rate features as well as subjective trust scores from self-assessment reports. The objective was to use rough set analysis to determine if there was a relationship between the measured objective features and self-reported trust scores, and the degree to which the objective features contributed towards trust estimation. A number of facial expression and voice features were found to be associated to subjective trust estimation.

Objective: Our objective was to examine the extent to which increasing precision of predictive (rate of change) information in process control will improve performance on a simulated process-control task. Background: Predictive displays have been found to be useful in process control (as well as aviation and maritime industries). However, authors of prior research have not examined the extent to which predictive value is increased by increasing predictor resolution, nor has such research tied potential improvements to changes in process control strategy. Method: Fifty nonprofessional participants each controlled a simulated chemical mixture process (honey mixer simulation) that simulated the operations found in process control. Participants in each of five groups controlled with either no predictor or a predictor ranging in the resolution of prediction of the process. Results: Increasing detail resolution generally increased the benefit of prediction over the control condition although not monotonically so. The best overall performance, combining quality and predictive ability, was obtained by the display of intermediate resolution. The two displays with the lowest resolution were clearly inferior. Conclusion: Predictors with higher resolution are of value but may trade off enhanced sensitivity to variable change (lower-resolution discrete state predictor) with smoother control action (higher-resolution continuous predictors). Application: The research provides guidelines to the process-control industry regarding displays that can most improve operator performance.