Humanizing Robot Mapping
Designing a human-in-the-loop interaction for map understanding in domestic robots
S.F. Gruben (TU Delft - Industrial Design Engineering)
J.H. Boyle – Graduation committee member (TU Delft - Materializing Futures)
O. Siebinga – Mentor (TU Delft - Materializing Futures)
More Info
expand_more
Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons.
Abstract
This graduation project explores how a user-in-the-loop approach can improve the usability and reliability of robot-generated maps in domestic environments. While Simultaneous Localization and Mapping (SLAM) allows robots to autonomously create maps of new environments, they are often imperfect due to sensor noise, environmental complexity, and localization errors. Such map errors can negatively impact navigation performance and reduce user trust, particularly when errors are difficult to interpret or correct.
The research phase focused on understanding how humans perceive and interact with robot-generated maps and identifying opportunities for a human-robot collaborative mapping approach. Through literature research and iterative robotics prototyping, insights were gathered and used to create a prototype interface that enables users to interpret, adjust, and refine robot-generated maps. The proposed concept views the user as an active collaborator in the mapping process, combining human spatial reasoning with algorithmic map generation. The design was evaluated through user testing, in which both map understanding and navigation performance were assessed.
The results show that users are able to identify and correct structural inconsistencies in robot-generated maps, leading to more visually consistent representations. Although no statistically significant improvements in navigation performance were found, the robot was able to successfully navigate using refined maps, and the results indicate a positive trend. This suggests that the approach is feasible and does not negatively impact system performance.
This project demonstrates that integrating users into the mapping process can enhance transparency and support more intuitive interactions with robotic systems. While further research is required to validate the effectiveness of the approach, the concept provides a promising direction for collaborative human-robot interaction, contributing to the development of more understandable and user-centered domestic robotics.