PhiCube: a reconfigurable bilateral robotic device for neurorehabilitation.

Introduction: Neuromotor disorders affecting upper-limb function represent a substantial clinical challenge in pediatric populations, with conditions such as hemiplegic cerebral palsy imposing lasting limitations on functional independence and quality of life. Bilateral motor training has emerged as a neurophysiologically grounded paradigm, offering functional advantages over purely unilateral approaches by actively exploiting interlimb coordination mechanisms. Despite the growing evidence base for bilateral training, existing robotic devices present critical limitations in kinematic reconfigurability, parameterizable inter-limb coupling, and portability, restricting their clinical accessibility and therapeutic versatility, particularly in pediatric settings. Methods: This paper introduces PhiCube, a modular robotic platform engineered for bilateral upper-limb neuromotor rehabilitation in pediatric populations. The device architecture is built around a compact central body housing two independent motorized rotational joints, whose actuation axes can be rapidly aligned with any of the three principal anatomical planes through a dedicated reconfiguration mechanism. A set of interchangeable therapeutic manipulanda, each eliciting distinct motor patterns, extends the device's therapeutic scope across the proximal-to-distal upper-limb kinematic chain. The bilateral coupling controller is formulated as an impedance-based scheme exposing tuneable parameters governing the kinematic transmission ratio between limbs, the directional asymmetry of the coupling, and the bilateral assistance level. A gamification environment embeds motor training within interactive paradigms designed for pediatric engagement, with real-time mapping between game trajectories and robotic assistance references. Kinematic simulations were conducted within the OpenSim framework, integrating an upper-extremity musculoskeletal model through a custom Python-based inverse kinematics pipeline. Results: Kinematic simulations confirmed that each handle-orientation combination elicits a distinct and characterizable pattern of joint recruitment across the upper-limb kinematic chain, with configurations ranging from isolated distal training (wrist flexion-extension, forearm pronation-supination) to compound multi-joint synergies engaging the shoulder, elbow, and wrist simultaneously. Analysis of the bilateral controller through contour mapping of the torque field demonstrated that the coupling parameters shape the inter-limb interaction, spanning configurations from unilateral guidance-in which one limb drives the other without reciprocal constraint-to fully symmetric bilateral coupling, and from in-phase synchronous to counter-phase alternating coordination. Discussion: The parametric characterization of both the kinematic architecture and the control space provides objective evidence that PhiCube can address a broad and clinically relevant spectrum of bilateral rehabilitation paradigms, from proximal multi-joint synergies to isolated distal training, and from strict guided-movement to symmetric and counter-phase bilateral coordination. These properties position PhiCube as a versatile tool for neuromotor rehabilitation, where therapeutic goals, residual motor capacity, and anthropometric constraints vary substantially across patients and evolve throughout the course of treatment.