| Summary: | The implementation of Fuzzy control systems on microcontrollers provides a low-cost, efficient, and innovative approach for assistive technologies, such as wheelchairs controlled by head and neck motion. However, their practical implementation is sensitive to several design parameters that are often overlooked. This study presents a statistical evaluation of a Fuzzy control system implemented on Arduino Mega and Due microcontrollers, focusing on the effects of the Iteration Step (IS) during defuzzification, microcontroller architecture, and overlap between membership functions, using the Center of Gravity (COG) method. System performance was evaluated using Root Mean Squared Error (RMSE) and processing time. The influence of an overlap between membership functions on the RMSE was also investigated. It was found that a smaller IS, such as 0.01, increased the precision of the system but led to long processing times (over 700 ms). Intermediate IS values (0.1 and 1) offered the most favourable trade-off between speed and accuracy, with RMSEs as low as 0.13% of the motor's total output and processing times under 260 ms. The Arduino Due significantly outperformed the Mega in processing time at low IS values, with little variation in accuracy. At higher ISs, the difference in processing time was negligible for practical use, making the Mega a viable cost-effective option. These findings provide empirical guidance for optimising Fuzzy control parameters in low-cost control systems, especially in applications where both real-time responsiveness and accuracy are critical. IMPLICATIONS FOR REHABILITATION: Optimising the operating parameters of fuzzy control systems is critical to enhancing the usability, responsiveness, and reliability of power wheelchairs for users with cervical mobility. Systems based on head and neck motion offer promising solutions for individuals with severe upper limb impairments, providing greater autonomy while remaining non-invasive and intuitive. By statistically evaluating performance across different microcontroller architectures and tuning parameters, this study contributes to the development of low-cost, accessible assistive technologies. The results offer practical guidance for implementing and optimising efficient fuzzy logic controllers in affordable wheelchairs, potentially broadening access to independent mobility solutions.
|
|---|