<![CDATA[Deaf-mute individuals face significant communication barriers due to limited public familiarity with sign language. In Indonesia, SIBI (Sistem Isyarat Bahasa Indonesia) is the government-standardised one-handed finger-spelling system used as the basis of communication for the hearing-impaired. This paper presents the design, implementation, and evaluation of an IoT-based hand sign language translator glove that recognises all 26 SIBI alphabet letters and displays the result on an Android application. The glove integrates five flex sensors for finger-bending detection, a GY-91 module (MPU-9250 + BMP280) for wrist orientation measurement, and a CD4051 analog multiplexer, all processed by a Wemos D1 Mini (ESP8266) microcontroller. Sensor readings are classified using a threshold-based decision method calibrated across three subjects. Classified letter data are transmitted via MQTT over Wi-Fi to a cloud broker and rendered in real time by an Android application. Experimental evaluation covers flex sensor resistance characterisation for all 26 SIBI letters, GY-91 gyroscope orientation profiling, multi-subject threshold calibration, end-to-end application display accuracy, and voltage measurement error percentage. Results confirm that the combined flex-sensor and gyroscope approach identifies SIBI alphabet letters with 90.00% end-to-end display accuracy and a voltage measurement error below 5%, indicating the preliminary feasibility of a low-cost, single-hand wearable IoT glove as an assistive sign-language communication aid, based on testing with a small number of participants. The system in its current form translates individual static SIBI alphabet letters only; it does not yet recognise dynamic gestures, whole words, or continuous sentence-level sign language.]]>
Copyrights © 2026
