A traveler holds the handle of BlindPilot and walks around tables and seated people toward an empty chair. The robot guides the traveler to the chair and indicates that it is on their right.
We propose an assistive robot, BlindPilot, which directly guides blind users to landmark objects using an intuitive handle.
We propose an assistive robot, BlindPilot, which directly guides blind users to landmark objects using an intuitive handle.
BlindPilot uses an RGB-D camera to detect the position of an empty chair and uses a LiDAR to build a 2D map of the surrounding area. BlindPilot then generates a path to the chair and moves along the generated path.
System diagram showing how an RGB camera detects an empty chair and depth sensing estimates its 3D position. LiDAR-based mapping supports planning a path around obstacles to the chair.
BlindPilot uses an RGB-D camera to detect the position of an empty chair and uses a LiDAR to build a 2D map of the surrounding area. BlindPilot then generates a path to the chair and moves along the generated path.
Side-by-side evaluation footage compares searching for a chair with audio navigation and a white cane against walking with BlindPilot. Overlaid floor plans trace the routes from the starting point to the chair; the footage is sped up.
Our user study showed that BlindPilot enabled users to approach a chair faster with a greater feeling of security and less effort compared to a sound-based local navigation system.
Our user study showed that BlindPilot enabled users to approach a chair faster with a greater feeling of security and less effort compared to a sound-based local navigation system.
30-sec Preview Video
BlindPilot guides a blind traveler around obstacles to an empty chair using a robotic handle.
Main Video [1:44]
BlindPilot guides a blind traveler around obstacles to an empty chair using a robotic handle.
Abstract
Blind people face various local navigation challenges in their daily lives such as identifying empty seats in crowded stations, navigating toward a seat, and stopping and sitting at the correct spot. Although voice navigation is a commonly used solution, it requires users to carefully follow frequent navigational sounds over short distances. Therefore, we presented an assistive robot, BlindPilot, which guides blind users to landmark objects using an intuitive handle. BlindPilot employs an RGB-D camera to detect the positions of target objects and uses LiDAR to build a 2D map of the surrounding area. On the basis of the sensing results, BlindPilot then generates a path to the object and guides the user safely. To evaluate our system, we also implemented a sound-based navigation system as a baseline system, and asked six blind participants to approach an empty chair using the two systems. We observed that BlindPilot enabled users to approach a chair faster with a greater feeling of security and less effort compared to the baseline system.
Publication
Seita Kayukawa, Tatsuya Ishihara, Hironobu Takagi, Shigeo Morishima, and Chieko Asakawa.2020.BlindPilot: A Robotic Local Navigation System that Leads Blind People to a Landmark Object.Extended Abstracts of the 2020 CHI Conference on Human Factors in Computing Systems (CHI EA 2020).
