ETD

• experimental validation
• intracortical brain-computer interface
• robotic upper-limb control
• ROS 2 platform
• spinal cord injury

Cervical spinal cord injuries (SCI) often cause tetraplegia, leading to severe motor and sensory impairment in all four limbs. Restoring upper-limb function is a primary goal for improving quality of life in people with SCI. Among available approaches, implantable brain–computer interfaces (iBCI) can decode motor intentions from the cortex and translate them into control commands for assistive devices.
This work, carried out with the Cortical Bionics Research Group (CBRG), presents the development and validation of a robotic platform integrating a dexterous arm–hand system, both controlled via intracortical signals recorded from the motor cortex of a participant with C4 SCI. The control algorithm was implemented in the Robot Operating System 2 (ROS 2) framework.
Neural activity recorded during a reach–grasp–carry–release task was analyzed to characterize functional selectivity and task-phase modulation, revealing distinct cortical activations across movement phases. Decoding was performed using a Regularized Inverse Optimal Linear Estimator (RIOLEd) model, reconstructing 6-DoF Cartesian arm velocities and a grasp velocity at 50 Hz.
A Cartesian velocity controller based on Damped Least Squares inverse kinematics converted decoded velocities into joint commands at 500 Hz. The temporal mismatch between decoder and controller rates was compensated through interpolation and update-rate adaptation. Hand control used kinematic remapping to transform grasp velocity into coordinated trajectories of the five fingers.
Validation under controlled synthetic inputs, offline neural integration, and real-robot execution demonstrated high decoding accuracy, with R², RMSE, and Pearson correlation confirming reliable trajectory tracking and effective real-time functional upper-limb control.