Neurorobotics: Brain-Machine Interface Arm Control (2D)

This 2D companion drives the same brain-machine interface decode pipeline as the 3D version through a plain canvas view built for reading the neuroscience rather than orbiting a scene: a ring of cosine-tuned cortical neurons fires proportionally to how closely its own preferred direction matches the intended reach, a population-vector decoder reconstructs that intended direction from the weighted sum of the population's noisy activity, and a 2-link planar arm carries out the decoded movement with a live raw-vs-smoothed decode-error readout. Adjusting the recorded neuron count, neural noise, tuning-curve sharpness and decode update rate shows exactly how population coding, sample size and temporal filtering combine to recover a clean movement signal from a noisy neural population — the same principles behind real motor-cortex BMI research such as the BrainGate trials.