Understanding Expectations for a Robotic Guide Dog for Visually Impaired People

TL;DR

This paper addresses the design of robotic guide dogs to assist visually impaired users by examining four design dimensions: locomotion, navigation movements, bidirectional communication, and explainability. A user study with 18 BVI participants compares gait controllers (model-based MPC versus learning-based RL), diverse navigation maneuvers, harness versus leash interactions, and audio explainability. The results favor learning-based locomotion with lower noise, gradual turning, rigid handles, and detailed, customizable audio explanations, underscoring the importance of personalization and context-aware behavior. These findings provide concrete design implications for creating practical, acceptable, and safe robotic guide dogs capable of operating across varied environments.

Abstract

Robotic guide dogs hold significant potential to enhance the autonomy and mobility of blind or visually impaired (BVI) individuals by offering universal assistance over unstructured terrains at affordable costs. However, the design of robotic guide dogs remains underexplored, particularly in systematic aspects such as gait controllers, navigation behaviors, interaction methods, and verbal explanations. Our study addresses this gap by conducting user studies with 18 BVI participants, comprising 15 cane users and three guide dog users. Participants interacted with a quadrupedal robot and provided both quantitative and qualitative feedback. Our study revealed several design implications, such as a preference for a learning-based controller and a rigid handle, gradual turns with asymmetric speeds, semantic communication methods, and explainability. The study also highlighted the importance of customization to support users with diverse backgrounds and preferences, along with practical concerns such as battery life, maintenance, and weather issues. These findings offer valuable insights and design implications for future research and development of robotic guide dogs.

Figures (6)

• Figure 1: Overview of the study setup.
• Figure 2: Comparison of ratings for two types of robot locomotion, Model Predictive Control (MPC, red) and Reinforcement Learning (RL, blue). (a) Noise (silent to overwhelming): MPC demonstrates a generally higher noise level than RL. (b) Compliance (resistive to compliant): RL shows wider variation. (c) Satisfaction (dissatisfied to satisfied): RL shows tighter clustering with a higher median, indicating more consistent satisfaction than MPC.
• Figure 3: Ratings of using leash handles on clarity$\uparrow$, security$\uparrow$, and frustration$\downarrow$, compared to rigid handles on a scale of -2 to +2. Overall, leash handles were rated lower in clarity and security and higher in frustration than rigid handles.
• Figure 4: Ratings of Modified PIADS (n=14) for the robotic guide dog without and with audio explanation. Overall, participants estimated a positive impact from the robotic guide dog alone, and greater when audio explanation is included, compared to their current mobility aids. Item i11 reverse coded.
• Figure 5: Robot and human movement diagram for Part 2.