Work

Edge-AI Enabled Intraoral Tongue Gesture Interface for Intelligent Robotic Control

Year: 2026
Type: conference-paper
Abstract: Assistive control technologies for individuals with severe motor impairments are often limited by high cost, hardware complexity, and reliance on external processing systems. This study proposes a compact and affordable device inside the mouth, and tongue gesture recognition interface designed for intelligent robotic control applications. The system uses four Force Sensitive Resistor(FSR) sensors within a custom made mouthguard to detect directional tongue presses corresponding to forward, backward, left, and right commands. Sensor signals are processed locally using an ESP32 microcontroller, where a K-Nearest Neighbour (KNN) classifier performs real-time edge-based gesture recognition without cloud dependency. Classified commands are transmitted via Bluetooth Low Energy (BLE) to operate an external robotic platform. The methodology includes optimized sensor placement, intraoral insulation design for safety and durability, analog signal conditioning, supervised data acquisition, feature extraction. And performance evaluation using classification accuracy, confusion matrix analysis, and latency measurement. A comparative analysis between conventional threshold-based control and machine learning classification will also be conducted to validate system robustness. The expected outcome is a reliable, low-latency assistive interface capable of achieving high real time accuracy while maintaining low implementation cost. All classification, latency and noise analyses were conducted in a MATLAB simulation environment, showing a proof-of-concept validation prior to physical hardware deployment. This work contributes to intelligent robotics and embedded AI by presenting a practical human and machine interaction system aligned with the themes of intelligent robotics and sustainable smart systems.
Language: English
Open Access status: hybrid
OpenAlex ID: W7204196004
FWCI:
Cites: 10
Cited by:
Related to: 10