Brain-computer interfaces allow users to interact with their environment through a paradigm such as motor imagery. The system reads the electroencephalographic signals, which are processed to interpret the user’s intention. This study uses the motor imagery paradigm to control a hand exoskeleton; the analysis of 10 young adults was done using the training data in a proprietary tool and an open software tool. Five feature extraction methods were used; three focused on spatial filtering techniques to enhance the separability of different mental states. These included Filter Bank Common Spatial Pattern, Filter Bank Common Spatio-Spectral Pattern, and Regularized Filter Bank Common Spatial Pattern. The other two algorithms, namely Band Log-Variance and Band Power Ratio, were energy and power distribution techniques that focused on analyzing those two variables across different frequency bands. The SelectPercentile algorithm was used to select the 10% most informative features, which were used to perform classification by logistic regression. The objective was to find out the most successful feature extraction technique. Filter Bank Common Spatio-Spectral Pattern yielded the best results, with an accuracy of 80.51% and an F1-Score of 82.4%.