Public evidence of ongoing validation.
We present Zenodo preprints alongside execution-based validation results. The full methodology and performance metrics are disclosed.
Preprint · nmFUSION
Holding the Edge:
Behavioral Evidence for Neuromorphic Threshold Management of ELM-like Instability in BOUT++
A neuromorphic controller was evaluated in the BOUT++ elm_pb plasma instability workflow. Across nine learned runs, eight delayed runaway onset by 44±2% relative to baseline, shifting the first runaway crossing from simulation time 50 to 72±6, and extended residence near the instability threshold before full solver stress emerged.
Preprint · nmCRYPT
Time-Born Entropy:
Neuromorphic Chaotic Dynamics as a Software-Defined Source of Cryptographic Randomness
A neuromorphic, chaotic, time-axis-driven software entropy source was evaluated across NIST STS, dieharder, and SP 800-90B. In the latest direct 90B run, the raw source reached H = 7.883983 bits/byte on the IID path and a final conservative non-IID value of 7.322342 bits/byte, supporting the possibility of software-defined temporal dynamics as a credible source of cryptographic randomness.
Preprint · SLNN Camouflage
Neuromorphic Active Camouflage:
Multiband Sensor Deception Through Adaptive Surface Metacontrol
A neuromorphic SLNN metacontrol architecture is proposed for multiband active camouflage across visible and infrared sensing. Rather than claiming perfect invisibility, the preprint frames camouflage as a measurable sensor-deception problem: reducing color and thermal contrast, delaying identification, and degrading tracking stability through adaptive tile-level control, surface feedback, and online recovery under contamination, damage, and heat-source variation.
Execution Evidence · nmVOR
Neuromorphic Visual Stabilization:
Retinal Slip Suppression Under Dynamic Disturbance
A neuromorphic controller was evaluated on vestibulo-ocular-reflex-style camera stabilization under continuous gyroscopic disturbance. Using pitch and roll motion, retinal slip, actuator velocity, and predicted next-step slip as inputs, it generates dual-axis torque commands that drive actuator angle and actuator velocity to hold gaze closer to stability than a conventional OIS baseline.
Inputs: gyro pitch/roll, retinal slip, actuator velocity, predicted slip
Outputs: dual-axis torque, actuator angle, actuator velocity
Comparison: retinal slip magnitude, residual velocity, counterphase alignment, cancellation efficiency
Execution Evidence · nmHAND
Neuromorphic Grip Control:
Slip Reduction With Lower Force Overshoot
A neuromorphic controller was evaluated on slip-aware robotic grasp control using tactile and inertial sensing only. From tactile slip, slip rate, grip force, grip velocity, micro-vibration, and lateral shear, it produces tighten, hold, and release effort that reduces tactile slip while avoiding the excess-force behavior of a reactive PID-style baseline.
Inputs: tactile slip, slip rate, grip force, grip velocity, micro-vibration, lateral shear
Outputs: tighten, hold, release, grip effort
Comparison: tactile slip, grip margin, overgrip, balance cost
