Brain-computer interface (BCI) controlled wheelchair robots can serve as powerful aids for severely disabled people in their daily life, especially to help them move voluntarily. In order to better understand human “thought”, owing to the development of the hybrid brain/muscle interface technique, in this paper, we present a real-time hybrid brain/muscle interface to control a wheelchair directly to keep the disables recovering several motion capabilities by using noninvasive motor imagery Electroencephalography (EEG) and Electromyography (EMG). The EMG and EEG signals from the users are extracted to control the motion of an intelligent wheelchair. Both signals processing consists of off-line training, online control evaluation, and real-time control. An algorithm called the common spatial patterns (CSP) is used in this human-robot system to extract the most discriminative spatial patterns pairs as features. The extensive experiments were conducted on the developed human-wheelchair systems to verify the proposed approaches.