For most of the last decade, the bottleneck in motor neuroscience
and rehabilitation was not theory but instrumentation — too few
channels, too much drift, too little dialogue between the cortex
and the machine that carried the body.
We design from the other side of that constraint: assume tens of
thousands of stable channels, assume sub-millisecond closed loops,
assume an exoskeleton that does not merely amplify intention but
negotiates with it. The work that follows is what such an
assumption demands.
Our practice is split between three benches and one shared
telemetry room. We publish the parts of our work that benefit
from being public, and we keep the rest in the lab where it
belongs. We are uninterested in the demo cycle.
Most of what we ship is invisible from the outside: jigs,
cabling, drivers, calibration rigs, training protocols, and
the patient years of arguing with our own data.