Over the last years evidence has accumulated that shows the possibility to analyze human brain activity on-line and translate brain states into actions such as selecting a letter from a virtual keyboard or moving a robotics device. These initial results have been obtained with either invasive approaches (requiring surgical implantation of electrodes) or synchronous protocols (where brain signals are time-locked to external cues). In this paper we describe a portable noninvasive brain-computer interface that allows the continuous control of a mobile robot in a house-like environment and also the operation of a virtual keyboard. The interface works asynchronously (the person makes self-paced decisions on when to switch from one mental task to the next) and uses 8 surface electrodes to measure electroencephalogram signals from which a statistical classifier recognizes 3 different mental states. Here we report results with five volunteers during their brain-actuated interaction experiments with the mobile robot and the virtual keyboard. Two of the participants successfully moved the robot between several rooms, while the other three participants managed to write messages with the virtual keyboard. One of the latter volunteers is a physically impaired person suffering from spinal muscular atrophy.