WanderGuide: Indoor Map-less Robotic Guide for Exploration by Blind People

TL;DR

The paper tackles the problem of enabling independent indoor exploration for blind people without relying on prebuilt maps. It introduces WanderGuide, a map-less robotic guide that combines autonomous navigation with real-time, multi-level scene descriptions generated by an MLLM, and supports Q&A and Take-Me-There interactions. Through a formative study and a main user study with blind participants, the authors identify three user-preference groups for description detail and demonstrate that users value wandering with guidance, while highlighting MLLM as a bottleneck for highly detailed, POI-specific information. The work advances accessible robotics by validating a map-less exploration paradigm, outlining design requirements, and proposing concrete system components and interaction modes, with implications for future enhancements and real-world deployment.

Abstract

Blind people have limited opportunities to explore an environment based on their interests. While existing navigation systems could provide them with surrounding information while navigating, they have limited scalability as they require preparing prebuilt maps. Thus, to develop a map-less robot that assists blind people in exploring, we first conducted a study with ten blind participants at a shopping mall and science museum to investigate the requirements of the system, which revealed the need for three levels of detail to describe the surroundings based on users' preferences. Then, we developed WanderGuide, with functionalities that allow users to adjust the level of detail in descriptions and verbally interact with the system to ask questions about the environment or to go to points of interest. The study with five blind participants revealed that WanderGuide could provide blind people with the enjoyable experience of wandering around without a specific destination in their minds.

Figures (5)

• Figure 1: Image of the robot and handle interface used in the study. Panel A-1 shows the robot used in the formative study, while Panel A-2 presents the robot used in the main study. Panels B-1 and B-2 illustrate the mapping of the handle interface buttons' functions, depending on the selected navigation mode.
• Figure 2: Floor maps of the location of the study. The left panel shows the two floors of the science museum, the fifth floor of Miraikan, which feature exhibits on various topics, such as environmental issues and space exploration. On the right panel is a floor plan of a shopping mall, the fourth floor of Toranomon Hills Station Tower, which includes a variety of restaurants offering different cuisines, including French, Japanese, Chinese, and cafes.
• Figure 3: Examples of descriptions described in the formative study. Panel A shows an example of a description generated at the science museum, and Panel B shows the one generated at a shopping mall.
• Figure 4: Three steps of the waypoint detection algorithm. Step 1 shows the generated cost map, while Step 2 depicts the skeletonization process of the cost map along with the detection of intersection points. Finally, Step 3 highlights the selected intersection points, which are identified as waypoint candidates.
• Figure 5: Box plot of evaluation with human experts in seven-point Likert points.