What My Eyes Can't See, A Robot Can Show Me: Exploring the Collaboration Between Blind People and Robots
Mayara Bonani, Raquel Oliveira, Filipa Correia, André Rodrigues, Tiago Guerreiro, Ana Paiva · 2018 · Proceedings of the 20th International ACM SIGACCESS Conference on Computers and Accessibility (ASSETS 2018) · doi:10.1145/3234695.3239330
Summary
This University of Lisbon and University of São Paulo study explores how assistive robots can go beyond information-giving technologies (like screen readers) by physically collaborating with blind people. The research consists of two studies. Study 1 conducted four focus groups with 20 blind screen reader users to understand how blind people perceive and envision robots in their daily lives. None had previously interacted with robots. Participants expressed common concerns about safety, over-reliance, reliability, cost, and security ("In a few years, a virus can turn off everything in your house"), but envisioned robots across a remarkably broad range of scenarios: navigation (from street-finding to beach towel locating), housekeeping (cooking, cleaning, laundry), education (teaching children colors and shapes, judging children's handwriting for blind parents), social companionship (dancing, playing games, chatting), personal servants (finding and fetching objects), and contextual information (identifying dropped items, checking food validity, evaluating clothing coordination). Study 2 compared two forms of robotic assistance in a Tangram puzzle assembly task with 12 blind participants (ages 21-64, mean 48) using a Baxter humanoid robot: Voice-Only Assistive Robot (VOAR, providing static verbal instructions) and Collaborative Assistive Robot (CAR, physically guiding participants' hands to pieces and target positions, with feedback adapted to their progress). Both conditions used Wizard of Oz methodology.
Key findings
The results were dramatic. In the CAR condition, all 12 participants (100%) successfully completed the Tangram puzzle. In the VOAR condition, only 2 participants (16.7%) completed it — a statistically significant difference (chi-squared = 17.143, p < 0.001). Average time to place each piece was significantly faster with CAR (63.9 seconds) than VOAR (202.2 seconds, p = 0.003). Participants perceived the CAR as significantly warmer (5.5 vs 4.6, p = 0.0), more competent (6.4 vs 5.7, p = 0.00), and more useful for the task (6.9 vs 5.0, p = 0.01) and for everyday life (5.7 vs 3.7, p = 0.0). They also reported higher positive feelings (6.3 vs 5.2, p = 0.02) and perceived the task as easier (1.9 vs 4.5, p = 0.01) with CAR. Discomfort levels did not differ between conditions. In debriefing interviews, all participants found Baxter useful (11 of 12) and likeable (all 12), with all but one preferring the CAR. Participants valued that the robot's mechanical sounds helped them understand its position and state (moving vs stationary). Eleven of 12 considered the robot's movement speed adequate and respectful of personal space. The CAR included a "handshake" introduction where participants could explore the robot's arms before starting — a design choice that helped establish the physical collaboration relationship.
Relevance
This study provides the first empirical evidence that physically collaborative robots dramatically outperform voice-only assistive technology for blind users in manipulation tasks. The 100% vs 16.7% task completion rate is a striking demonstration that verbal instructions alone are fundamentally insufficient for guiding spatial assembly — blind users need physical guidance that robots, unlike smartphones and screen readers, can uniquely provide. For the accessibility field, this suggests that robots could fill a critical gap between information-only assistive technology and the physical assistance currently provided only by sighted peers. The focus group findings reveal that blind people envision robots far beyond navigation (the focus of most assistive robotics research) — household tasks, parenting support, education, and social companionship were all prominently desired. The social perception findings are also significant: physical collaboration made participants perceive the robot as warmer and more competent, suggesting that embodied physical interaction may be key to building trust and positive relationships between blind users and assistive robots.
Tags: blindness · robotics · human-robot interaction · assistive robotics · collaboration · haptic technology · independence · Wizard of Oz