What is the ROBO-GAP dataset?

The ROBO-GAP (humanoid ROBOts - Gender and Age Perception) dataset is a collection of scores concerning the perceived age and gender (femininity, masculinity, and gender neutrality) attributed to the 251 robots currently featured in the ABOT (Anthropomorphic roBOT) dataset. To find out more about the research behind this database, watch our presentation at HRI 2022

How to cite the ROBO-GAP dataset?

Perugia, G., Guidi, S., Bicchi, M., & Parlangeli, O. (2022, March). The Shape of Our Bias: Perceived Age and Gender in the Humanoid Robots of the ABOT Database. In Proceedings of the 2022 17th ACM/IEEE International Conference on Human-Robot Interaction. (110-119) ACM/IEEE. [link]

You are free to use the data which can be downloaded from this cite for your study provided that you always credit the source and cite our paper.

What can you find in the ROBO-GAP dataset?

The ROBO-GAP dataset features the Mean, Standard Deviation, and bootstrapped 95% Confidence Intervals of the age, masculinity, femininity, and gender neutrality ratings attributed to each of the robots in the ABOT database, as well as the labels masculine, feminine, or neutral based on the combination of the different gender ratings.

What can you find on this website?

Besides allowing to Download the ROBO-GAP dataset, this website gives the opportunity to Explore the relationships between the age and gender ratings in the ROBO-GAP dataset using interactive scatterplots , and to view, filter and explore the dataset in tabular form in the Data View section. From this section you can even draw a random sample of robots from the whole dataset or a custom selection of it, and download the gender and age data as .csv file.

Find a robot

Select dimensions to plot

How to use

Click on any point in the plot to display information about the corresponding robot.

Find a specific robot, and display the average age score and gender ratings, select its name from the dropdown menu in the 'Find a robot' search box, and click on the 'Highlight' button.

Configure the plot, by selecting the ROBO-GAP variables to map to the x and y axes, and to the dots color.

Filter the robots in the plot by different variables using the sliders.

Draw a random sample

Choose the sample size and click the "Draw sample" button to select a random sample from the current selection. Please note that any filter applied from the filter area is considered before drawing the sample.

Download as .csv

Click on the "Download" button to save the data about the selected robots as .csv file.


Please remember to always cite our paper if you use the ROBO-GAP data in a study or elsewhere.

© Giulia Perugia, Stefano Guidi, Margherita Bicchi, Oronzo Parlangeli - all rights reserved.

Download the dataset

You can donwload the data in .csv format at following [link]

How was the data collected?

The robots in the ABOT dataset were randomly divided into five groups. 153 participants were randomly allocated to one of the five groups and evaluated the perceived age (on a scale from 0 to 100) and gender (on a 7-point Likert scale) of 50-51 robots. The pictures of the robots were presented in randomized order, and were kept in their original size. Each robot in the ROBO-GAP dataset was rated by at least 25 participants.

How was the data analyzed?

For each robot we computed the mean and standard deviation of the ratings across observers, for all the rated dimensions. The reliability of the aggregated ratings for each dimension was assessed using a version of Conger’s Intraclass Correlation Coefficient , ICC(2, k), which ranged from .896 for gender neutrality to .954 for femininity (age: ICC(2, k) = .929, masculinity: ICC(2, k) = .932), indicating excellent reliability. 95% confidence intervals for the mean ratings for each dimension were then computed using the bootstrap method (N = 1000).

Do you want to replicate the ROBO-GAP dataset?

Since the dataset was collected only with Italian participants. We encourage replication of this study with participants from other countries. Please, reach out to us in case you plan to replicate our study or have already done so.

Watch our presentation at HRI 2022!

You can watch the presentation given by Giulia Perugia at HRI 2022 about the ROBO-GAP study and dataset in the video below.

Meet the ROBO-GAP team!

The ROBO-GAP study was conceived and conducted by a team of researcher from the University of Siena and the Eindhoven University of Technology. Below you can find a short profile of all the members of the team.

Giulia Perugia

Giulia Perugia, PhD received a M.Sc. in Cognitive Science from the University of Siena, and a double degree Ph.D. in Assistive Technologies from Eindhoven University of Technology (TU/e, Netherlands) and the Technical University of Catalonia (UPC, Spain). She worked as a postdoctoral researcher at Uppsala Social Robotics lab (Sweden) from 2018 to 2021. Currently, she is an Assistant Professor at the Human-Technology Interaction group of TU/e. She is interested in studying how people’s perceptions and affective states in the interaction with robots can be measured through the interaction itself, how such states and perceptions develop over time, and whether and how the design of and interaction with social robots could reproduce biases and stereotypes existing in society at large. Recently, she has started investigating more profusely the process of gendering robots, how this interacts with robot’s norm violations, and how it could be used to foster a more inclusive HRI.

Stefano Guidi

Stefano Guidi, PhD, is Assistant Professor of psychology at the University of Siena, Department of Social, Political and Cognitive Sciences. He is currently Professor of General Psychology and of Methodlogy of Research in Cognitive and Social Psychology. At the University of Siena is member of the Work Environment Laboratory. Prof. Guidi’s research interests range from visual perception (contour integration, visual completion and experimental visual aesthetics) to social cognition (age, gender and sexual orientation stereotypes in response to human and artificial agents), moral judgements and work-related stress. A specific focus of his research is on research integrity in relationship with psychosocial working conditions. He is also interested in quantitative research methods in psychological science. He has collaborated on national and European research projects.

Oronzo Parlangeli

Oronzo Parlangeli, PhD, is Associate Professor of psychology at the University of Siena, Department of Social, Political and Cognitive Sciences. He is currently Professor of Cognitive Psychology and Cognition, Technology and Sustainability. At the University of Siena he is also the scientific director of the Work Environment Laboratory and the president of the Ethics Committee for Research in Human and Social Sciences. From 2007 to 2010 he was National President of the Italian Society of Ergonomics. Prof. Parlangeli’s main research interests concern the study of cognitive processes, such as perceptual and learning processes, and of the relationships between them and cognitive ergonomics with particular reference to contexts such as health, training, safety, social media, and social robotics. A specific focus of his research is on research integrity. He has directed and collaborated on national and European research projects.

Margherita Bicchi

Margherita Bicchi, BS, graduated in Communication Sciences at the University of Siena, where she is now studying for her master’s degree in Communication Strategies and Techniques. Her university courses are mainly concerned with the study of Design Thinking, User Experience and Digital Communication. She is interested in the psychological, social and ethical issues of Human-Robot Interaction.