Envisioned Utility of BMIs

The full extent to which BMIs would impact human behavior is vastly unknown. Yet, short-term possibilities are innumerable. For example, there is a growing consensus that BMIs could provide the only viable short-term therapeutic alternative to restore motor functions in patients suffering from extensive body paralysis (including lack of communication skills) resulting from devastating neurological disorders.

Assuming that noninvasive techniques to extract large-scale brain activity with enough spatial and temporal resolution can be implemented, BMIs could also lead to a major paradigm shift in the way normal healthy subjects can interact with their environment. Indeed, one can envision a series of applications that may lead to unprecedented ability to augment perception and performance in almost all human activities. These applications would involve interactions with either real or virtual environments. According to this view, real environments can also include local or remote control relative to the human subject, while virtual environments can be realistic or intentionally unrealistic. Here are some of examples.

1. Local, real environment: Restoration of the motor function in a quadriplegic patient. Using a neurochip implanted in the subject's brain, neural signals from healthy motor brain areas can be used to control an exoskeletal or prosthetic robotic arm used to restore fundamental motor functions such as reaching, grabbing, and walking.

2. Remote, real environment: Superhuman performance, such as clearing heavy debris by a robot controlled by the brain signals of a human operator located far away from the danger zone. Recent results by the P.I. and his collaborators have demonstrated that such remote control could be achieved even across the Internet.

3. Realistic virtual environment: Training to learn a complex sequence of repair operations by the trainee's brain directly interacting with a virtual reality program, with or without the involvement of the trainee's peripheral sensorimotor system.

4. Unrealistic virtual environment: Experiencing unrealistic physics through a virtual reality system for a "what if" scenario, in order to understand deeply the consequences of terrestrial physics.

Given the significant degree of plasticity documented even in the adult brain, repeated use of BMIs will likely transform the brain itself, perhaps more rapidly and extensively than what is currently possible with traditional forms of learning. For example, if a robot located locally or remotely is repeatedly activated via a BMI, it is likely that cortical areas specifically devoted to representing the robot will emerge, causing the robot to effectively become an extra limb of the user.

What real advantages might we obtain from future BMI based devices, compared to more conventional interfaces such as joysticks, mice, keyboards, voice recognition systems, and so forth? Three possible application domains emerge:

1. Scaling of position and motion, so that a "slave" actuator, being controlled directly by the subject's voluntary brain activity, can operate within workspaces that are either far smaller (e.g., nanoscale) or far bigger (e.g., space robots; industrial robots, cranes, etc.) than our normal reach

2. Scaling of forces and power, so that extremely delicate (e.g., microsurgery) or high-force tasks (e.g., lifting and displacing a tank) can be accomplished

3. Scaling of time, so that tasks can be accomplished much more rapidly than normal human reaction time, and normally impossible tasks become possible (e.g., braking a vehicle to a stop after seeing brake lights ahead; catching a fly in your hand; catching something you have dropped; responding in hand-to-hand combat at a rate far exceeding that of an opponent)

To some extent, all these tasks, with the exception of time scaling, can, in principle, be accomplished though conventional teleoperator systems in which the human using his limbs operates a master device, which, in turn, controls a local or remote slave device. There is a history of five decades of research in this area of robotics, with moderate success, such as recent commercial development of teleoperated surgical systems. Major difficulties have been the design of appropriate master devices that the human can interact with naturally and the destabilizing effects of long time delay between the master and the slave. BMIs offer unique advantages in two ways:

1. They eliminate the need for master devices that interact with the human

2. Since the human is directly operating through his brain, the time delays associated with the signal transmission from the peripheral sensors to the CNS (~ 10-30 msec) and from CNS to the muscles (~ 10-30 msec), and then the time required of a limb to complete the needed action (~100-900 msec), can be reduced by an order of magnitude.

Elimination of the need for a master device is a radical departure from conventional teleoperation. Furthermore, the reduction of time delays leads to the exciting possibility of superhuman performance. For example, moving an arm from point A to point B can take ~500 msec from the time muscles are commanded by the brain, because of the force generation limitations of the muscles, the inertia of the arm, and the need to accelerate from A and to decelerate to B. But if a slave robot that is much better than the human arm in terms of power/mass ratio is directly controlled though a BMI, all three types of time delays (peripheral sensory, motor signal transmission, and limb motion) can be minimized or eliminated, possibly leading to faster and more stable operation of the slave robot. For instance, it is possible for an impaired or unimpaired person to wear an arm exoskeleton that directly interacts with the brain much faster than the natural arms.

In recent years, work developed by our laboratories has demonstrated the feasibility of building BMIs dedicated to the task of utilizing brain-derived signals to control the 1-D and 3-D movements of artificial devices. In a series of studies, we have provided the first demonstrations in animals that such BMIs can be built, that animals can learn to operate these devices in order to obtain a reward, and that motor control signals derived from the extracellular activity of relatively small populations of cortical neurons (50-100 cells) can be used to reproduce complex 3-D arm movements in a robotic device in real time.

Recent advances in nanotechnology could help significantly the advance of this area of research. First, this technology could provide new ways to extract large-scale brain activity by reducing the degree of invasiveness of current electrophysiological methods. Investment on research aimed at designing a new generation of VLSI aimed at both conditioning and analyzing large-scale electrical brain activity will also be required. Finally, a complete new generation of actuators, designed to operate in micro- or nanospaces needs to be built, since there are many new applications that can be envisioned if brain-derived signals can be employed to directly control nanomachines.

0 0

Post a comment