Prototype: Portable Braille Glove

In both medical and everyday settings, individuals who are deaf-blind face significant communication barriers that can compromise their autonomy, safety, and access to critical information [8]. Moreover, communication with deaf or hard-of-hearing individuals, especially through traditional methods such as tactile sign language or interpreter-based support, is often constrained by the limited availability of trained personnel and the need for direct physical interaction. These limitations are particularly problematic in urgent or dynamic environments, where timely and effective communication is essential [3].

To address these challenges, this project proposes a haptic glove designed to translate text, and eventually speech, into real-time tactile feedback using a Braille-inspired pulse pattern [2][9]. By delivering information directly to the hand through strategically positioned actuators, the glove provides a discreet, fast, and independent communication channel for deaf-blind users [1]. This technology has the potential to significantly enhance accessibility, particularly in healthcare contexts where rapid, private communication is vital.

The design of the glove is grounded in research on the sensitivity of different areas of the hand [1][5]. Actuators are positioned accordingly to maximize tactile perception while maintaining the standard six-dot Braille layout [4]. To adapt to real-time spoken communication, traditional Braille punctuation symbols are repurposed to represent common environmental sounds or auditory cues (e.g., alarms, traffic). This enables the system to relay ambient auditory information using tactile means, enhancing situational awareness for deaf-blind individuals [7].

Importantly, the relevance of including deaf-blind individuals in accessible communication solutions is growing, with an estimated 0.2–2% of the global population experiencing some form of deaf-blindness or direct interaction with affected individuals [6]. Despite this, technological support remains limited. By emphasizing usability, responsiveness, and portability, the proposed haptic glove presents a practical and inclusive solution [10]. Unlike conventional methods that rely on interpreters or fixed setups, this wearable device empowers users with independent access to information on the go. Its portable design supports continuous use across various environments, making it a versatile and impactful alternative to traditional communication aids.

To successfully realize this project, the following components are required to build a functional prototype. Additionally, a detailed bill of materials, including pricing as of the date of publication, is provided in the attached Excel file.

1. 1x Arduino micro
2. 6x Titan TacHammer
3. 2x breadboard
4. 60x jumper wires
5. 1x Arduino Micro USB Cable
6. 1x Multiplexer
7. 6x Three-Port Terminal
8. 6x DRV2605L Haptic Motor Controller
9. 1x Arduino bluetooth wireless transceiver
10. 1x 9V battery
11. 1x battery clip
12. 1x PC with Python software

To implement this project, the Arduino software is required and can be downloaded from the following link: https://www.arduino.cc/en/software. This software enables you to upload all necessary code to the Arduino successfully.

The system is built into a glove equipped with six vibration actuators, each representing one dot in a standard 6-dot Braille cell. Spoken input is captured through the computer's microphone, transcribed into text using a speech recognition engine, and then converted into Braille patterns via a Python script. These patterns are sent to an Arduino, which activates the actuators to physically render each character through tactile feedback.

Glove Design

Six coin vibration motors are mounted corresponding to the six dots of a standard Braille cell onto the back of a glove, allowing participants to continue using their hands for everyday tasks, such as grasping objects, without interference during or between tests [2][9]. According to previous research on tactile sensitivity and spatial discrimination of the hand, the most suitable locations for actuator placement, maximizing distinguishability and comfort, are primarily on the proximal phalanges (backs of the fingers) and the upper palm [1][4]. Each motor was securely attached using adhesive and connected with flexible silicone wiring, which was carefully routed and stitched into the glove fabric to maintain comfort and reduce cable strain during hand movements.

To build the prototype of the Braille Glove, both the Python code and the Arduino code (provided in the attachment) are required. Below is a brief explanation of how each part of the system works:

Python-side processing (PC)

1. Speech Recognition

Spoken input is captured using the computer’s microphone and converted to text using the Google Speech Recognition API through the speech_recognition Python library.

2. Braille Mapping

Each character is translated into a 6-bit Braille pattern, represented as binary values (0–63). Each bit corresponds to one of the six Braille dots.

3. Serial Communication

The binary Braille pattern is transmitted to the Arduino Micro over USB using the pyserial library. Each character is followed by a short pause, and spaces are marked with a dedicated 'S' signal.

Arduino-side Processing (Embedded System)

4. Receiving & Parsing

The Arduino reads incoming serial data and interprets each line as either a Braille pattern or a space indicator.

5. Multiplexer Switching

Since all DRV2605L haptic drivers share the same I²C address, a TCA9548A I²C multiplexer is used to assign each actuator to a separate I²C channel (1–6).

6. Haptic Feedback Generation

For each character, the Arduino activates the corresponding actuators (for bits set to ‘1’) and sends a short PWM pulse to the DRV2605L driver, generating tactile feedback.

For our test, we worked with a group of 20 participants (family and friends) who had no prior experience with Braille. They were only provided with a Braille alphabet chart (see image above) to help interpret the signals. In addition, one participant with limited vision and prior knowledge of Braille was included in the test group to compare with.

Graph 1: Effect of inter-pulse duration on perception

In this evaluation, participants were presented with a series of 4- to 6-letter words in random order. For each successive word, the interval between pulses was decreased by 125 milliseconds. The test concluded when a participant failed to correctly identify a word, resulting in a failure score assigned to that pause duration and all shorter intervals. The same group of participants completed two additional sessions, each conducted at least five hours apart from the previous one.

The participants were presented with the same set of words each time, but in a different random order for each trial and participant. This was done to ensure that participants could not predict which word would come next, preventing any learning effects or anticipation during the test.

The words used included:

1. Hello
2. Arm
3. Haptic
4. Braille
5. Yours
6. Bye
7. Could
8. Humans
9. Vision
10. To

Graph 2: Clarity of different words

This test focused on evaluating the varying difficulty levels associated with different words, in order to identify design pain points. A pulse interval of 1.0 seconds was used, based on the results of the previous experiment. All participants received the same set of words, which were provided in advance, as well as the Braille alphabet. For each word, participants were asked to rate how clearly it was perceived through the Braille glove, expressed as a percentage. The complete results are available in the attached Excel file, and the average ratings are visualized in the graph above.

The words used:

1. Hello
2. nice
3. to
4. meet
5. you
6. haptic
7. interfaces

Graph 1: Effect of inter-pulse duration on perception

The graph illustrates the percentage of participants (out of a sample of 20) who correctly understood a word transmitted through a series of pulses, with varying pauses between each pulse. The x-axis represents the length of the pause in milliseconds, while the y-axis shows the percentage of participants who correctly identified the word.

There is a clear downward trend in recognition performance as the pause between pulses becomes shorter. This suggests that shorter pauses make the signal more difficult to interpret.

On the first try, performance is high when pauses are long (≥1750 ms), with all participants correctly identifying the word. However, accuracy drops rapidly as the pause decreases below 1375 ms, reaching zero when the interval falls below 750 ms. With additional attempts (the second and third tries), participants generally improved at understanding the signal. This suggests a learning effect, where participants become better at interpreting the signal with practice.

Between the second and third attempts, there is a clear improvement in the shortest pause duration at which 100% accuracy is maintained. While full accuracy in the second test was only achieved at pauses of around 1500 ms and longer, the third test shows that all participants were able to correctly interpret the word even at shorter pauses, down to approximately 1250 ms. This indicates that, even after the initial improvement, participants continued to refine their skills with further exposure. The progression from the second to the third test highlights how repeated practice can fine-tune perception, enabling accurate recognition under increasingly time-constrained conditions.

The purple line represents a single participant with prior experience in reading Braille. This individual was able to correctly interpret the signals on the first try, even with pauses as short as 625 ms between letters, significantly outperforming the rest of the group. This highlights the strong impact of prior training or familiarity on the ability to process and understand pulse-based signals effectively.

This graph demonstrates that longer pauses between pulses lead to higher word recognition rates. It also shows that training or repeated attempts improve performance, and prior experience greatly enhances signal comprehension. These results underline the importance of adequate spacing between pulses for clear communication and the benefits of familiarization with the signal format.

Graph 2: Clarity of different words

The graph provides a clear representation of the varying levels of difficulty associated with Braille word recognition. A general trend observed is that word length correlates negatively with recognition accuracy — longer words tend to be more difficult to identify. This effect can be attributed to the participants’ lack of prior experience with Braille. As evidenced by the first test results, increased familiarity with Braille correlates positively with improved recognition performance.

An additional finding is the facilitative effect of letter repetition within a word. Words containing repeated characters were more readily recognized, likely due to the reinforcement of identical pulse patterns. For instance, the words “hello” and “meet” received higher accuracy scores compared to other words of similar length, presumably because of the repeated l and e characters, respectively.

In contrast, the word “you” exhibited notably low recognition scores relative to other three- and four-letter words. This appears to stem from the letter “y”, which is encoded using a complex five-dot Braille pattern. This denser pulse combination likely introduced greater cognitive load and perceptual difficulty, leading to a significant drop in recognition accuracy.

In addition, the word “interfaces” yielded the lowest recognition score among all tested words. This aligns with the previously observed trend that recognition accuracy decreases as word length increases. “Interfaces”, being the longest word in the test set, likely imposed a higher cognitive and perceptual load on participants, particularly due to the greater number of distinct Braille characters and pulse sequences involved. For individuals without prior Braille experience, sustaining attention and accurately interpreting a long sequence of unfamiliar tactile signals becomes increasingly challenging.

While the system demonstrates strong potential, improvements in actuator resolution, adaptive timing algorithms, and user training modules could further enhance usability. Future studies should explore long-term adaptation among users with visual and auditory impairments, particularly in real-world, high-noise environments.

This project successfully demonstrated the feasibility of a wearable haptic glove that translates spoken language into tactile Braille-inspired signals. By integrating speech recognition, text processing, and precise haptic feedback, the system offers a discreet, portable communication tool for individuals who are deaf-blind.

The glove’s most impactful feature is its independence from interpreters or fixed setups, enabling real-time communication in dynamic environments like hospitals or transit systems. Its lightweight design allows continuous use without limiting hand function.

Key findings:

1. Higher recognition accuracy with longer pulse intervals
2. Rapid learning effect through repeated exposure and learning
3. Increased difficulty in recognizing longer words or complex pulse signals

Challenges for the future

1. Feedback optimization to improve clarity
2. Creating a standalone, user-friendly interface, so there is no need for a PC
3. Adaptive timing could personalize pulse speeds based on user ability or word complexity, helping new users process signals more comfortably.
4. Developing effective training tools for new users

This project highlights the power of user-centered design in assistive technology. Even a simple tactile interface, when designed thoughtfully, can offer meaningful independence. We hope this work lays the foundation for further innovation in tactile communication and inspires others to advance accessible, inclusive technology.

[1] Kilbom, C. F.-H. (1993, June 3). Sensitivity of the hand to surface pressure. Opgehaald van ScienceDirect: https://www.sciencedirect.com/science/article/pii/000368709390006U?via%3Dihub

[2] Tanay Choudhary, S. K. (2015, January 8). A Braille-based mobile communication and translation glove for deaf-blind people. Opgehaald van IEEE Xplore: https://ieeexplore.ieee.org/abstract/document/7087033/figures

[3] Mann, T. H. (2007, June 14). Adapting Tests of Sign Language Assessment for Other Sign Languages—A Review of Linguistic, Cultural, and Psychometric Problems. Journal of deaf studies and deaf education, 138 - 147. Opgehaald van Oxford academic: https://academic.oup.com/jdsde/article/13/1/138/497542?login=true

[4] Dahiya, O. O. (2022). Smart Tactile Gloves for Haptic Interaction,Communication, and Rehabilitation. Glasgow: Wiley-VCH GmbH. Opgehaald van https://advanced.onlinelibrary.wiley.com/doi/epdf/10.1002/aisy.202100091

[5] Johansson, J. L. (1983, October 4). Regional differences and interindividual variability in sensitivity to vibration in the glabrous skin of the human hand. Brain Research - ScienceDirect, 65 - 72. Opgehaald van ScienceDirect: https://www.sciencedirect.com/science/article/pii/0006899384904037

[6] Luz, T. R. (2024, February). Caeski: an assistive technology for the communication of persons with deafblindness. Disability and rehabilitation: Assistive technology, 281 - 291. Opgehaald van KU Leuven Limo: https://www-tandfonline-com.kuleuven.e-bronnen.be/doi/full/10.1080/17483107.2022.2087768#d1e301

[7] Nathan R. Luzum, B., Benjamin L. Hamel, M., Valeriy Shafiro, P., & Michael S. Harris, M. (2023, September). Identification Accuracy of Safety-Relevant Environmental Sounds in Adult Cochlear Implant Users. The Laryngoscope, 2388 - 2393. Opgehaald van https://onlinelibrary.wiley.com/doi/epdf/10.1002/lary.30475

[8] Wolsey, J.-L. A. (2017, July 8). Perspectives and Experiences of DeafBlind College Students. Opgehaald van The Qualitative Report: https://nsuworks.nova.edu/tqr/vol22/iss8/1/

[9] M. Rajasenathipathi, M. A. (2010, September 3). An electronic design of a low cost Braille handglove. International Journal of Advanced Computer Science and Applications (IJACSA), 52 - 57. Opgehaald van https://www.proquest.com/docview/2656846225?parentSessionId=5xbn0sZPyepUEJ9QflIzKW8fWkx086vBC3%2B3O1OgytY%3D&pq-origsite=primo&accountid=17215&sourcetype=Scholarly%20Journals

[10] Lenin R. Villarreal, B. J. (2020). Wireless Haptic Glove for Interpretation and Communication of Deafblind People. In P. H.-C. Vanessa Agredo-Delgado, Human-Computer Interaction (pp. 305 - 314). Arequipa: Springer. Opgehaald van https://link-springer-com.kuleuven.e-bronnen.be/chapter/10.1007/978-3-030-66919-5_31

This short reel demonstrates how our haptic glove prototype uses tactile feedback to support real-time communication for individuals who are deaf-blind. By converting spoken language into Braille-inspired vibrations on the hand, the glove provides an independent and portable solution to a critical medical communication challenge. Watch as the system captures speech, processes it into text, and delivers clear, tactile signals, making accessibility more immediate and hands-on.