One of the main reasons that simple neural sampling as used in
these devices will not work is already quite clear, and stems
from basic anatomy. While the motor cortex, which is being
sampled in these devices, outputs the base movement, this
output does NOT drive the musculature directly, but rather is
routed through the cerebellum, where the actual movement is
fine tuned and coordinated with overall body kinetics. To
acquire smooth, coordinated movement instructions from a
neural sample, cerebellar input must be sampled. ASAICT it
currently is not. This is not surprising, as the cerebellar signal
is highly complex, and deeply encoded.
As the use of prosthetic limbs increases in military veterans, the Pentagon is investigating prostheses that are more durable, reliable and directly controlled using brain implants.
DARPA, the military’s research arm, said it will launch the next phase of its decade-old Revolutionizing Prosthetics program, which had an original goal to create a fully-functioning, neurally-controlled human limb within five years.
Though the agency has made considerable progress — human trials of the DEKA Arm are underway, and a neurally-controlled arm is under development at Johns Hopkins University — it hasn’t yet achieved its goal.
The hurdles:
- It has proved difficult to fully integrate human neural pathways with artificial platforms.
- Neural-recording interfaces have short life spans of just two years.
- Neural-recording interfaces don’t extract adequate information to yield seamless movement from brain to neurons to limbs.
- Current prototypes can’t move fast enough: even at 500 events per second, it’s not enough for fluid motion.
To face the challenge, DARPA is launching its Histology for Interface Stability Over Time program.
The goal: create a neurally-controlled limb that lasts for 70 years and has complete integration with the human body.
Here’s what the agency says (.pdf):
DARPA is soliciting innovative research proposals in the area of neural-recording interface failure analysis. The HIST program seeks to develop the technology needed to reliably extract information from the nervous system, and to do so at a scale and rate necessary to control many degree-of-freedom (DOF) machines, such as high-performance prosthetic limbs. Technologies and techniques emerging from this program will enable the construction of reliable neural-recording interfaces, which will be suitable for clinical use over the lifetime of an injured soldier (~70 years). Additionally, an objective understanding of the failure mechanisms will lead to high-throughput biological testing, due to the discovery of predictive markers linked to a high probability of failure and other accelerated-testing techniques. Proposed research should investigate innovative approaches that enable revolutionary advances in science, devices, or systems. Specifically excluded is research that primarily results in evolutionary improvements to the existing state of practice.
In other words: DARPA wants to understand why neural-recording interfaces are so unreliable, and how failure can be predicted before an amputee is left without the use of an artificial limb.
The program is structures in three phases over three years. It’s basically like a hacker contest for prosthetic limbs — DARPA wants researchers to overload neural systems to find vulnerabilities.
Of particular concern are “implanted cortical microelectrodes,” or brain implants, which DARPA believes may be the best system for the job.
Images: U.S. Army; Johns Hopkins University Applied Physics Laboratory
