Motorized Mini Exercise Bikes (MMEBs), have found applications in Brain Computer Interfaces (BCIs) for rehabilitation, aiming to enhance neural plasticity and restore limb movements. However, processing electroencephalography (EEG) data in this context presents challenges, often relying on discrete on/off control strategies. Such limitations can impact rehabilitation progress and Human-Machine Interaction (HMI). This study introduces a Support Vector Machine (SVM)-based approach to classify passive pedaling tasks at varying speeds using EEG signals. The research protocol involved four healthy volunteers performing passive pedaling induced by a MMEB at two speeds: 30 and 60 rpm. SVM achieved an average ACC of 0.77, a false positive rate of 0.26, and AUC of 0.80, demonstrating the feasibility of accurately identifying passive pedaling at both low and high speeds using EEG signals. These results hold promising implications for improving the design of more robust and adaptive controllers in BCI systems integrated with MMEBs, particularly within the context of lower limb rehabilitation. This research supports the way for enhanced brain-machine interaction, offering potential benefits to individuals with disabilities by facilitating more precise control of rehabilitation devices and advancing the field of neuroengineering. Further exploration of real-world applications and broader implications is necessary to fully harness the potential of this SVM-based approach.
