Bi-directional control of supernumerary limbs

Performance site: Campus Biotech, Geneva

Background

Supernumerary fingers and limbs devices are now emerging as a tool for augmentation of motor functions. While wearable extra fingers are primarily envisioned for disabled people usage, supernumerary limbs generally aim at enhancing able-bodied people abilities,  (Parietti, Chan et al. 2014, Wu and Asada 2015, Hussain, Salvietti et al. 2017). In our group, we focus on the latter category, considering three aspects of such device.

First, a primary challenge for such devices is to allow the user to concurrently and independently coordinate such human-machine interface (HMI) with the overt movements of the intact limbs. In this perspective, the recent results of Bashford and collaborators are compelling. They showed the feasibility to simultaneously control a brain-computer interface (BCI) and natural finger movements independently (Bashford, Wu et al. 2018).

Second, we consider the role of sensory feedback to enhance the experience with a supernumerary arm.  Evidence on the need for haptic feedback for improved motor control of supernumerary artificial limbs are emerging (Hussain, Meli et al. 2015, Parietti and Asada 2017), and the pivotal role in HMI and BCI has been largely demonstrated (D’Anna, Petrini, et al. 2017, Corbet, Iturrate, et al. 2018, Petrini, Valle, et al. 2019).

Finally, we aim at understanding the sense of embodiment and agency when a subject interacts with supernumerary arms. The embodiment of a tool into a person’s body schema is widely reported in cognitive neuroscience (Maravita and Iriki 2004, Sengül, van Elk, et al. 2012). The feeling of independent ownership of a third virtual hand with no dis-ownership of the real ones was demonstrated by Bashford and colleagues (Bashford and Mehring 2016). As pointed out by the authors, understanding whether an independent neural representation of a supernumerary limb could be generated as a consequence of its real control through an HMI is an intriguing and open issue. Also, studies have shown that force feedback enhances robotic tool integration (Sengül, Rognini et al. 2013). However, it is so far unknown the extent to which haptic sensory information plays a role in the creation of ownership and agency feelings when dealing with the control of a supernumerary arm.

Project description: 

Have you ever seen someone who can voluntary move his/her ears? Extrinsic auricular muscles (AMs) – namely the posterior, superior, and anterior auricular muscles (PAM, SAM, AAM) – helps to pivot the ears in multiple directions towards sounds of interest in mammals. In humans, they are considered vestigial [1], and their original role in the humans’ pinna orienting system is not functional anymore, even though weak attention-related activations have been reported [3].

Nevertheless, those who do not naturally have this ability can learn it through training [2]. Differently from non-human species, AMs’ cortical representation in humans is located in the primary motor cortex, suggesting evolution in functionality and the possibility to contract AMs at will [4]. It has been also shown that subjects can learn to modulate two different activity’s bandwidths of the SAM superficial electromyogram (sEMG), achieving proficient two-dimensional cursor-to-target control [5]. In addition, it has been shown that neuromuscular electrical stimulation (NMES) could facilitate learning in people with no initial voluntary control of PAM [6].

Therefore, AMs activation can be considered a fruitful source of signal for human-machine interface (HMI) control and has been exploited for wheelchair and prosthesis control [2, 7].

Here, we consider AMs as a suitable source of control for HMI for human augmentation. As a final goal, healthy subjects will be trained to control in a coordinated fashion three arms.

AM vestigial nature makes it possible to repurpose them without impairing existing motor abilities. On the other hand, the activity of adjacent muscles is unlikely to interfere and producing sEMG artifacts. Eventually, the location of AMs and in particular of the PAM constitutes a reasonably favorable recording site in terms of signal quality, device encumbrance for the subject, and cosmetics [8].

Activities:

In light of the above, we are planning to achieve control over the end effector (hand) position of a supernumerary arm using AMs sEMG signals. To this aim, the student will be asked to:

  • Identify the most suitable of the AM muscles as well as optimal decoding approach for the control of a point in space;
  • Design a training paradigm for subjects to learn adequate volitional control of the given AM;
  • Evaluate the outcomes of the combination of control strategy and subjects’ training.

Requirements:

Project is 30% hardware, 40% software, 30% experimentation

Knowledge in programming with Matlab and data analysis

Best for master projects

Contact: [email protected]

References:

[1]         M. Liugan, M. Zhang, and Y. O. Cakmak, “Neuroprosthetics for Auricular Muscles: Neural Networks and Clinical Aspects,” Frontiers in Neurology, 10.3389/fneur.2017.00752 vol. 8, p. 752, 2018.

[2]         L. Schmalfuß et al., “Steer by ear: Myoelectric auricular control of powered wheelchairs for individuals with spinal cord injury,” Restorative Neurology and Neuroscience, vol. 34, pp. 79-95, 2016, doi: 10.3233/RNN-150579.

[3]         S. A. Hackley, “Evidence for a vestigial pinna-orienting system in humans,” Psychophysiology, vol. 52, no. 10, pp. 1263-1270, 2015/10/01 2015, doi: 10.1111/psyp.12501.

[4]         J. Meincke, M. Hewitt, M. Reischl, R. Rupp, C. Schmidt-Samoa, and D. Liebetanz, “Cortical representation of auricular muscles in humans: A robot-controlled TMS mapping and fMRI study,” PLOS ONE, vol. 13, no. 7, p. e0201277, 2018, , : 10.1371/journal.pone.0201277.

[5]         C. Perez-Maldonado, A. S. Wexler, and S. S. Joshi, “Two-Dimensional Cursor-to-Target Control From Single Muscle Site sEMG Signals,” IEEE Transactions on Neural Systems and Rehabilitation Engineering, vol. 18, no. 2, pp. 203-209, 2010, doi: 10.1109/TNSRE.2009.2039394.

[6]         S. Chanthaphun, S. L. Heck, C. J. Winstein, and L. Baker, “Development of a training paradigm for voluntary control of the peri-auricular muscles: a feasibility study,” Journal of NeuroEngineering and Rehabilitation, vol. 16, no. 1, p. 75, 2019/06/14 2019, doi: 10.1186/s12984-019-0540-x.

[7]         L. Schmalfuss et al., “A hybrid auricular control system: direct, simultaneous, and proportional myoelectric control of two degrees of freedom in prosthetic hands,” Journal of Neural Engineering, vol. 15, no. 5, p. 056028, 2018/08/23 2018, doi: 10.1088/1741-2552/aad727.

[8]         R. N. Friedman, R. G. McMillan, C. J. Kincaid, and M. R. Buschbacher, “Preliminary Electrophysiological Characterization of Functionally Vestigial Muscles of the Head: Potential for Command Signaling,” The Journal of Spinal Cord Medicine, vol. 22, no. 3, pp. 167-172, 1999/01/01 1999, doi: 10.1080/10790268.1999.11719566.