Robotic assistive devices, such as exoskeletons are used in labour environments to promote social inclusion of diverse types of impairments as for example upper- limb. Robotic exoskeletons can be controlled by surface electromyography signals. However, people with severe neural impairments and absence of residual muscu- lar activity are unable of using these sEMG-based systems due to the absence of residual muscular activity. Alternatively, robotic hand prostheses and exoskeletons commanded by Brain-Computer Interfaces (BCIs) have been successfully applied in these people. This study aims to develop a low-cost steady-state visual evoked potential (SSVEP)-based BCI for social inclusion, using unsupervised calibration. A low-cost flicker visual stimulator with geometric shapes is proposed to elicit brain commands. Both Canonical Correlation Analysis (CCA) and Power Spectral Den- sity (PSD) are used to classify SSVEP stimuli. As a first step, the proposed BCI was tested in a serious game, which was developed to simulate the workspace, and provide feedback to the subject. CCA presented the best classification results with an accuracy of 71.6 ± 9.7% and an Information Transfer Rate (ITR) of 37.6 ± 15.4 bits/min and averaged latency of 0.77 ± 0.39 s to provide an output associated to the stimulus observed by the subject.