A Database for Kitchen Objects: Investigating Danger Perception in the Context of Human-Robot Interaction
Repository for information and discussion of the hri-object dataset: https://hri-objects.leusmann.io/
The dataset includes 153 Kitchen Objects (Images and Danger Ratings for three different scenarios).
Read the paper.
For questions and discussion please use and create an issue in this repository or contact: jan.leusmann@ifi.lmu.de
If you use the dataset please cite our work:
@inproceedings{leusmann2023database,
title = {A Database for Kitchen Objects: Investigating Danger Perception in the Context of Human-Robot Interaction },
author = {Jan Leusmann and Carl Oechsner and Johanna Prinz and Robin Welsch and Sven Mayer},
url = {https://payload.leusmann.io/publications/chiea23-577.pdf},
doi = {10.1145/3544549.3585884},
year = {2023},
date = {2023-04-23},
urldate = {2023-04-23},
booktitle = {Extended Abstracts of the 2023 CHI Conference on Human Factors in Computing Systems},
publisher = {Association for Computing Machinery},
address = {New York, NY, USA},
series = {CHI EA\'23},
pubstate = {forthcoming},
tppubtype = {inproceedings}
}
In the future, humans collaborating closely with cobots in everyday tasks will require handing each other objects. So far, researchers have optimized human-robot collaboration concerning measures such as trust, safety, and enjoyment. However, as the objects themselves influence these measures, we need to investigate how humans perceive the danger level of objects. Thus, we created a database of 153 kitchen objects and conducted an online survey (N=300) investigating their perceived danger level. We found that (1) humans perceive kitchen objects vastly differently, (2) the object-holder has a strong effect on the danger perception, and (3) prior user knowledge increases the perceived danger of robots handling those objects. This shows that future human-robot collaboration studies must investigate different objects for a holistic image. We contribute a wiki-like open-source database to allow others to study predefined danger scenarios and eventually build object-aware systems.