Brain-Computer Interface (BCI) has developed a technology with considerable potential, enabling direct communication and control between the human brain and external devices. The present study aims to examine the viability and effectiveness of BCI frequency domain features with the help of three different classification methods for the healthy human person. In this context, electroencephalography (EEG) data was acquired from 10 healthy persons exhibiting diverse levels of control possibility. The participants engaged in training sessions to become familiar with the Brain-Computer Interface (BCI) system and acquire the skill of controlling a robotic arm via the use of motor imagery technique. In this context, ten frequency domain features were extracted to classify the left and right-hand movements for offline robotic arm control. After frequency domain features extraction, four different classification algorithms namely efficient logistic regression (ELR), ensemble boosted tree (EBT), and wide neural network (WNN) were critically compared in terms of classification accuracy. The results demonstrated that the ELR classifier yielded the highest accuracy, followed by the WNN and EBT classifier. Subsequently, these intentions were translated into control instructions that were then executed by the robotic arm.