We’re growing the reflex robot hands don’t have.

A living reflex for
the last millimetre.

A grasp can change with a slight shift in friction, angle or pressure. Senxe is exploring whether living neural circuits can help robot hands adapt at contact, within limits enforced by software.

Explore the architecture

Living neurons, on a short leash.

Schematic

Five steps to
a living reflex.

  1. Gate 0 · NextTissue
    viable
  2. Gate 1Loop
    closes
  3. Gate 2Tissue
    contributes
  4. Gate 3Matches
    baseline
  5. Gate 4Beats
    baseline

Can the tissue stay alive on the array? It has to survive on the electrodes and answer the same stimulation the same way, again and again.

Does the loop close? We stimulate the tissue, read its response and feed it back into the controller, reliably, run after run. Passing Gates 0 and 1 is our first milestone.

Is the tissue really contributing? Performance has to drop when we swap in dead tissue, or a scrambled, shuffled or frozen signal. This is the first real test of the tissue.

Can it match the baseline? With tissue in the loop, the robot has to do as well as its rule-based controller alone, on the same task, under the same conditions.

Can it beat the baseline? Faster learning, more robustness and lower energy use than the baseline.

Let’s build the next reflex. We’re exploring living neural systems for robotic control. Connect with us to discuss research collaborations, robotics applications, or early-stage investment.

Get in touch