Feedback interfaces

None of the myoelectric prostheses currently used in clinical practice have purposely designed closed-loop controllers. Therefore, all current SoA feedback interfaces are still in the laboratory development phase, with the exception of a rather simple one integrated in the Vincent Evolution 2 Prosthesis (Vincent Systems GmbH [34]). Even without its commercial counterpart, prosthetic feedback is a very relevant research topic as summarized in the expert-review article by Antfolk et al. [35]. There are many pathways available to close the loop between the user and the prosthesis (Figure 1.8);

but the prevalent one is the non-invasive (i.e., cutaneous). Methods for delivering

cutaneous feedback fall into two groups: modality-matched or sensory substitution.

Figure 1.8: Three possible feedback information pathways in the context of myoelectric prostheses. Different colors correspond to different pathways (A, B, C). Pathway A is related to sensory information that is directly fed back to the CNS (e.g., visual and auditory feedback); Pathway B to the information that is conveyed to functional sensory motor systems invasively or noninvasively; Pathway C is related to the intrinsic feedback. Image adapted from [35].

1.2.2.1 Modality matched SoA

Feedback is modality-matched when the output stimulus is felt in the same modality as the sensory input (e.g., temperature sensation is not substituted but rather directly transmitted by warming/cooling the skin surface). The development of non-invasive modality-matched feedback comes along with an array of unique challenges. In theory, it is possible to regain modality-matched touch sensations (contact, normal and shear force/pressure, vibration, texture, temperature) using noninvasive electromechanical devices coupled with thermoelectric ones (Figure 1.9). In their work, Kim et al. [36]

and Armiger et al. [37] presented a miniature SoA haptic device capable to transmit touch, pressure, vibration, shear force, and temperature to the skin of the user.

Perhaps the most challenging sensory input to replace is a proprioception. Here the joint angle (e.g., of the elbow) needs to be transferred to another unaffected joint, in order to match the modality. One possible solution to this problem was proposed by Goodwin et al. [38] and later exploited by Roll et al. [39]. They demonstrated that when vibration of around 80 Hz frequency and sufficient intensity is applied over the tendons at the wrist, the subjects perceive it as a joint motion. This phenomenon could be utilized to

transmit the prosthesis position in a modality matched way. Therefore, contrary to their intuitiveness, the modality-matched interfaces remain secondary to sensory substitution;

crucial obstacles for their successful implementation remain their size, interface, and power consumption.

Figure 1.9: a) Mechanical- and b) thermal-tactor used for modality matched feedback.

Images adapted from [37].

1.2.2.2 Sensory substitution SoA

Sensory substitution is a method to provide sensory information to the body either via a different sensory channel or by maintaining the same channel but by changing the modality. Typical examples of this include the substitution of vision with touch (e.g., Braille alphabet for visually impaired people) or of pressure with vibration. Its main drawback is the danger that the mapping between the physical variable and its representation could be unintuitive to the user. Even though it is not ideal, this technique has the virtue of a relatively straightforward implementation, which is the induction of either mechanical vibration (vibro-tactile) or electric current into the surface of the skin (electro-tactile) (see Figure 1.10).

The vibro-tactile stimulation is elicited on the surface of the skin by mechanical vibrations of the actuator or its contact tip. First such devices, developed specifically for prosthesis application and introduced in the early 50s, were quite bulky and power consuming [40]. But over time, the technology was perfected and they were made much more compact and energy efficient [41], [42]. Vibrotactile feedback activates mainly two types of mechanoreceptors in the skin: Pacinian corpuscles which react best to frequencies between 200 and 300 Hz, and Meissner corpuscles which are best activated by frequencies around 50 Hz [43]. The sensitivity to amplitude changes is highly dependent on the location. The detection threshold is lowest on the fingertips (0.07 µm at 200 Hz) and highest on the abdomen and the gluteal region (4–14 µm at

200 Hz) [44]. Until now, vibro-tactile feedback has been used for feeding back a variety of prosthesis states to the user. Some of the more noticeable applications include transmission of prosthesis force, velocity, aperture or elbow position [45], [46], [47], [48] .

Figure 1.10: a) The SoA C2-Tactor can be used for vibro-tactile stimulation (Engineering Acoustics, Inc, Florida, USA [42]); b) a disposable surface electrode typically used for electro-tactile stimulation (Spes Medica Genoa, Italy))

Electro-tactile stimulation induces an electric current originating from a surface electrode (e.g., typically gold, platinum, silver, or stainless steel) that passes through the skin and directly stimulates afferent fibers [49]–[51]. The current polarity and the size of the electrode determine how deep it penetrates the skin surface. This influences the type of sensory afferents activated, since the four types of mechanoreceptors are located at different depths of the dermal tissue [52]. The resulting sensation can be perceived as tingling, itching, buzzing and pinching as well as sharp, needle-like pain. In summary, the parameters of the stimulation (current, frequency, and pulse width) play an important role as do material, type, and size of the electrode, its placement location, and skin impedance. Even though initial research was conducted already in the early 70s [53], the application of the electro-tactile stimulation interface was delayed until the early 80s, primarily due to the interference to the EMG signals. Nowadays this pitfall is successfully resolved by using time or frequency division multiplexing [54], [55].

Similar to their vibro counterparts, the electro-tactile devices have been used in a variety of studies as the interface of choice that communicates prosthesis’ grip, finger force, or touch [56], [57], [58].

Overall, the two interfaces are functionally very similar and the choice between the one or the other is driven by practical considerations such as power consumption or available space and psychological implications - amputees that suffered from an electrical shock might be negatively predisposed towards electro-tactile stimulation.