High School Thesis: Voice-Controlled Wheelchair Design
Budget: $10 – $30 USD
I want to write a thesis for my high school project:
This thesis presents the design and implementation of an offline voice-controlled wheelchair system using an embedded edge-computing approach. The proposed system aims to improve accessibility for users with severe motor impairments by eliminating the need for conventional physical control interfaces. Voice commands are processed locally to ensure low latency, reliability, and safe operation without dependence on internet connectivity.
The system is built around the ESP32-S3-Korvo-2 development board and employs the ESP-SR (Skainet) framework for wake-word detection and speech command recognition. An integrated Acoustic Front End (AFE) performs noise suppression, echo cancellation, and MFCC feature extraction to enable robust speech recognition in noisy environments. The firmware is developed in C using ESP-IDF and FreeRTOS, implementing a finite state machine that prioritizes safety-critical commands and manages real-time motor control.
Directional commands are translated into PWM signals to drive DC motors through an L298N H-bridge, using a scaled robotic platform to emulate wheelchair motion. The system is validated through functional and performance testing, demonstrating reliable command recognition, offline operation, and end-to-end latency within safety limits.
The results confirm that low-cost embedded hardware can effectively support advanced speech-based control for assistive mobility devices, highlighting the potential of Edge AI in assistive robotics and cyber-physical systems.
This thesis presents the design and implementation of an offline voice-controlled wheelchair system using an embedded edge-computing approach. The proposed system aims to improve accessibility for users with severe motor impairments by eliminating the need for conventional physical control interfaces. Voice commands are processed locally to ensure low latency, reliability, and safe operation without dependence on internet connectivity.
The system is built around the ESP32-S3-Korvo-2 development board and employs the ESP-SR (Skainet) framework for wake-word detection and speech command recognition. An integrated Acoustic Front End (AFE) performs noise suppression, echo cancellation, and MFCC feature extraction to enable robust speech recognition in noisy environments. The firmware is developed in C using ESP-IDF and FreeRTOS, implementing a finite state machine that prioritizes safety-critical commands and manages real-time motor control.
Directional commands are translated into PWM signals to drive DC motors through an L298N H-bridge, using a scaled robotic platform to emulate wheelchair motion. The system is validated through functional and performance testing, demonstrating reliable command recognition, offline operation, and end-to-end latency within safety limits.
The results confirm that low-cost embedded hardware can effectively support advanced speech-based control for assistive mobility devices, highlighting the potential of Edge AI in assistive robotics and cyber-physical systems.