The upper limbs are essential for a variety of daily activities involving reaching, grasping, and manipulating objects. In individuals who have undergone transradial amputation, the loss of the hand and part of the forearm compromises the ability to perform actions that enable interaction with the environment and leads to significant functional compensations.

Myoelectric prostheses are designed to restore these abilities by decoding user movement intentions based on muscle activity measured from the residual limb. Despite major technological advances, their adoption remains limited, partly due to the lack of physiological sensory feedback, which is essential for closing the sensorimotor loop. Essentially, users need to be able to feel their robotic hand. Deprived of proprioceptive and tactile feedback, they rely heavily on their vision, experience increased cognitive load, and must undergo long and demanding training. These elements compromise movement accuracy and prosthesis acceptance. Consequently, artificial sensory feedback is attracting interest within the field of robotic prosthetics research. The majority of studies have focused on tactile feedback. However, proprioception, which is essential for motor control and the development of a sense of autonomy, remains underrepresented in the literature. Current approaches range from sensory substitution, such as vibrotactile stimulation, to biomimetic strategies, aiming at reproducing the original sensation.

This project proposes an intermediate bioinspired approach, between sensory substitution and biomimicry, based on the use of Transcutaneous Electrical Nerve Stimulation (TENS) to transmit proprioceptive information related to hand configuration. Stimulation is applied via surface electrodes to interact with the median and ulnar nerves, inducing sensations downstream of the stimulation site and perceived in the amputated hand. This strategy provides spatially coherent proprioceptive feedback while avoiding the need for surgical intervention.

Achieving the objectives of this project has allowed for the gradual evolution of human-machine interaction between users and the robotic hand. An initial study focused on developing a proof of concept to evaluate the ability of the feedback strategy developed to transmit relevant proprioceptive information to the user of a robotic hand. Participants consistently identified finger apertures conveyed via median or ulnar nerve stimulation and grasp types conveyed through concurrent stimulation of both nerves, with respective classification accuracies of 96.5% and 88.3%.