Brain-computer interfaces are developing rapidly. A common, non-invasive form is to measure EEG activity on the surface of the scalp and get the participants to direct their attention to different visual stimuli that correspond to different commands (e.g. up, down, left, right). A much more invasive method is to implant microelectrode arrays directly into the cortex. This allows a more direct measure of motor cortex activity, which is much more accurate.

It’s possible that encasing the entire brain in a set of microelectrode arrays would allow control over this robot’s motor functions as well as allowing sensory integration, but I don’t think getting the robot to work is the primary cause of Prof Panda’s concern.