Touching on elements for a non-invasive sensory feedback system for use in a prosthetic hand

Hand amputation results in the loss of motor and sensory functions, impacting activities of daily life and quality of life. Commercially available prosthetic hands restore the motor function but lack sensory feedback, which is crucial to receive information about the prosthesis state in real-time when interacting with the external environment. As a supplement to the missing sensory feedback, the amputee needs to rely on visual and audio cues to operate the prosthetic hand, which can be mentally demanding. This thesis revolves around finding potential solutions to contribute to an intuitive non-invasive sensory feedback system that could be cognitively less burdensome and enhance the sense of embodiment (the feeling that an artificial limb belongs to one’s own body), increasing acceptance of wearing a prosthesis.

A sensory feedback system contains sensors to detect signals applied to the prosthetics. The signals are encoded via signal processing to resemble the detected sensation delivered by actuators on the skin.

There is a challenge in implementing commercial sensors in a prosthetic finger. Due to the prosthetic finger’s curvature and the fact that some prosthetic hands use a covering rubber glove, the sensor response would be inaccurate. This thesis shows that a pneumatic touch sensor integrated into a rubber glove eliminates these errors. This sensor provides a consistent reading independent of the incident angle of stimulus, has a sensitivity of 0.82 kPa/N, a hysteresis error of 2.39±0.17%, and a linearity error of 2.95±0.40%.

For intuitive tactile stimulation, it has been suggested that the feedback stimulus should be modality-matched with the intention to provide a sensation that can be easily associated with the real touch on the prosthetic hand, e.g., pressure on the prosthetic finger should provide pressure on the residual limb. A stimulus should also be spatially matched (e.g., position, size, and shape). Electrotactile stimulation has the ability to provide various sensations due to it having several adjustable parameters. Therefore, this type of stimulus is a good candidate for discrimination of textures. A microphone can detect texture-elicited vibrations to be processed, and by varying, e.g., the median frequency of the electrical stimulation, the signal can be presented on the skin. Participants in a study using electrotactile feedback showed a median accuracy of 85% in differentiating between four textures.

During active exploration, electrotactile and vibrotactile feedback provide spatially matched modality stimulations, providing continuous feedback and providing a displaced sensation or a sensation dispatched on a larger area. Evaluating commonly used stimulation modalities using the Rubber Hand Illusion, modalities which resemble the intended sensation provide a more vivid illusion of ownership for the rubber hand.

For a potentially more intuitive sensory feedback, the stimulation can be somatotopically matched, where the stimulus is experienced as being applied on a site corresponding to their missing hand. This is possible for amputees who experience referred sensation on their residual stump. However, not all amputees experience referred sensations. Nonetheless, after a structured training period, it is possible to learn to associate touch with specific fingers, and the effect persisted after two weeks. This effect was evaluated on participants with intact limbs, so it remains to evaluate this effect for amputees.

In conclusion, this thesis proposes suggestions on sensory feedback systems that could be helpful in future prosthetic hands to (1) reduce their complexity and (2) enhance the sense of body ownership to enhance the overall sense of embodiment as an addition to an intuitive control system.