UbiBraille: Designing and Evaluating a Vibrotactile Braille-Reading Device
Hugo Nicolau, João Guerreiro, Tiago Guerreiro, Luís Carriço · 2013 · Proceedings of the 15th International ACM SIGACCESS Conference on Computers and Accessibility (Assets '13) · doi:10.1145/2513383.2513437
Summary
This paper presents UbiBraille, a wearable vibrotactile device that enables blind users to read textual information privately and inconspicuously by leveraging their existing knowledge of Braille. The device consists of six small vibration motors attached to adjustable aluminum rings worn on the index, middle, and ring fingers of both hands — directly corresponding to the six dots of a Braille cell and mirroring the finger positions used when writing with a Perkins Braille typewriter. When a character is transmitted, the motors corresponding to the raised dots of that character vibrate simultaneously, creating a mnemonic mapping between Braille writing chords and reading vibrations. This approach addresses two key limitations of current output modalities for blind users: audio (via screen readers or TTS) is not always appropriate due to noisy environments or privacy concerns, and refreshable Braille displays are expensive (up to ,000) and difficult for users with reduced tactile sensitivity (e.g., from diabetes). The hardware prototype uses lily pad vibe boards connected to an Arduino Mega ADK board. Two user studies were conducted with blind participants recruited from a formation centre for blind people in Portugal.
Key findings
In Study 1 (character recognition), 11 blind participants (ages 21-62) achieved an average character recognition accuracy of 82% (SD=17.25%) with only 10 minutes of warm-up training. Performance was highly character- and user-dependent: two participants achieved 100% accuracy while one reached only 54%. Letters requiring more than 3 dots and stimulation on both hands (N, O, V, Y, Z) were hardest to recognize, with Y having the highest error rate at 54.5%. More than half of errors (51.6%) involved misidentifying a single dot — either a false positive (feeling a vibration that wasn't there from adjacent finger crosstalk) or false negative (missing a vibration). Critically, participants' performance on standard Braille reading and writing tasks was significantly correlated with UbiBraille accuracy (Spearman rho=-.627, p=.039 for writing speed), confirming that Braille knowledge transfers to vibrotactile reading. Age, self-ratings, and digit span (working memory) were not significantly correlated with performance. In Study 2 (word reading), 7 participants read 5-character Portuguese words at four timing conditions. At the 2000ms condition (2 seconds per character), word recognition reached 88.57%, with the best participants achieving ~1 character per second at the 1000ms condition with 90-100% accuracy — equivalent to approximately 12 words per minute. A preliminary sentence-reading assessment with two participants showed one achieving a perfect score on 10 sentences and the other correctly identifying 8 of 10.
Relevance
UbiBraille addresses an important gap in assistive technology: the need for private, silent text output for blind users. While audio feedback via screen readers is the dominant modality, it is unsuitable in meetings, classrooms, noisy environments, or any situation requiring discretion. The device is particularly promising for deaf-blind individuals for whom both audio and visual channels are unavailable. The design choice to map vibrations to the same fingers used in Braille writing is elegant — it leverages years of existing muscle memory and Braille expertise rather than requiring users to learn an entirely new tactile encoding. The finding that Braille proficiency directly predicts UbiBraille performance validates this mnemonic approach. For practitioners, the work highlights that vibrotactile communication is feasible with commodity hardware (Arduino boards, small vibration motors), making the approach reproducible and affordable. The word and sentence reading results demonstrate that the technique can scale beyond character-level discrimination to practical communication, though reading speeds of 12 WPM are modest compared to audio output. The device is best positioned as a complement to, rather than replacement for, audio feedback — filling the specific niche where audio is inappropriate.
Tags: braille · vibrotactile · wearable technology · blind users · haptic feedback · assistive technology · privacy · deafblindness · communication · Arduino · text entry