Abstract
Controlling a prosthetic upper limb requires the reconstruction of multiple distal articulations. Moreover, the higher the amputation level, the more joints need to be reconstructed, and the less kinetic information is available in the residual limb. By exploiting contextual information, such as the position and orientation of a target in a reaching task, we aim to reconstruct the natural dynamics of the distal joints using recurrent neural networks. We compare performances of two models, an Echo State Network (ESN) and an LSTM, on two conditions: training on individual subjects, and training on a 5-fold CV on 15 subjects. We explored hyperparameters on both models: the ESN shows better performances on the single-subject task, and the LSTM shows better performances on the multiple-subject task. When looking qualitatively at the predictions, we observe that even if networks don’t have the same MSE errors, they perform the task well and are able to reach the targets most of the time. We further analyze the performance of the models on the multi-subject task and report different kinds of generalizations.
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Notes
- 1.
This was used to speed up the computations compared to a 10 or 20% connectivity that is usually used. In preliminary testings, we didn’t see significant changes in performance of the reservoir connectivity for this task.
- 2.
Each LSTM recurrent cell has 4 parameters: cell state, input gate, output gate and forget gate.
References
Chollet, F., et al.: Keras (2015). https://keras.io
Cimolato, A., Driessen, J.J., Mattos, L.S., De Momi, E., Laffranchi, M., De Michieli, L.: EMG-driven control in lower limb prostheses: a topic-based systematic review. J. NeuroEng. Rehabil. 19(1) (2022). https://doi.org/10.1186/s12984-022-01019-1
Farina, D., et al.: The extraction of neural information from the surface EMG for the control of upper-limb prostheses: emerging avenues and challenges. IEEE Trans. Neural Syst. Rehabil. Eng. 22(4), 797–809 (2014)
Hinaut, X., Trouvain, N.: Which hype for my new task? Hints and random search for reservoir computing hyperparameters. In: ICANN 2021 - 30th International Conference on Artificial Neural Networks, Bratislava, Slovakia, September 2021. https://inria.hal.science/hal-03203318
Hochreiter, S., Schmidhuber, J.: Long short-term memory. Neural Comput. 9(8), 1735–1780 (1997)
Jaeger, H.: The “echo state” approach to analysing and training recurrent neural networks-with an erratum note. Bonn, Germany: German National Research Center for Information Technology GMD Technical Report 148(34), 13 (2001)
Kingma, D.P., Ba, J.: Adam: a method for stochastic optimization (2017). https://arxiv.org/abs/1412.6980
Latash, M.L., Scholz, J.P., Schöner, G.: Toward a new theory of motor synergies. Mot. Control 11(3), 276–308 (2007)
Manceron, P.: IKPy: an inverse kinematics library aiming performance and modularity (v3. 3.3). Phylliade/ikpy (2022)
Mick, S., et al.: Shoulder kinematics plus contextual target information enable control of multiple distal joints of a simulated prosthetic arm and hand. J. Neuroeng. Rehabil. 18, 1–17 (2021)
Prahm, C., Schulz, A., Paaßen, B., Aszmann, O., Hammer, B., Dorffner, G.: Echo state networks as novel approach for low-cost myoelectric control. In: ten Teije, A., Popow, C., Holmes, J.H., Sacchi, L. (eds.) AIME 2017. LNCS (LNAI), vol. 10259, pp. 338–342. Springer, Cham (2017). https://doi.org/10.1007/978-3-319-59758-4_40
Saleh, Q.M.: Design of a Neuromemristive Echo State Network Architecture. Rochester Institute of Technology (2015)
Salminger, S., et al.: Current rates of prosthetic usage in upper-limb amputees – have innovations had an impact on device acceptance? Disabil. Rehabil. 44(14), 3708–3713 (2022). https://doi.org/10.1080/09638288.2020.1866684, pMID: 33377803
Ségas, E.: Contrôle biomimétique de prothèses à partir des mouvements résiduels et d’informations contextuelles. Theses, Université de Bordeaux, March 2023. https://theses.hal.science/tel-04164931
Segas, E., Mick, S., Leconte, V., Klotz, R., Cattaert, D., de Rugy, A.: Data and code for intuitive movement-based prosthesis control in virtual reality, October 2022. https://doi.org/10.5281/zenodo.7187851
Soechting, J.F., Lacquaniti, F.: Invariant characteristics of a pointing movement in man. J. Neurosci. 1(7), 710–720 (1981)
Tanaka, G., et al.: Recent advances in physical reservoir computing: a review. Neural Netw. 115, 100–123 (2019)
Trouvain, N., Hinaut, X.: reservoirpy: A Simple and Flexible Reservoir Computing Tool in Python, June 2022. https://inria.hal.science/hal-03699931
Acknowledgment
This work was funded by the ANR-DGA-ASTRID grant CoBioPro (ANR-20-ASTR-0012-1). We thank Aymar De Rugy, Effie Segas, Vincent Leconte and Bianca Lento from the INCIA HYBRID team for sharing data and collaborating on this project.
Author information
Authors and Affiliations
Corresponding authors
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2024 The Author(s), under exclusive license to Springer Nature Switzerland AG
About this paper
Cite this paper
Bernard, P., Alexandre, F., Hinaut, X. (2024). Prediction of Reaching Movements with Target Information Towards Trans-humeral Prosthesis Control Using Reservoir Computing and LSTMs. In: Wand, M., Malinovská, K., Schmidhuber, J., Tetko, I.V. (eds) Artificial Neural Networks and Machine Learning – ICANN 2024. ICANN 2024. Lecture Notes in Computer Science, vol 15025. Springer, Cham. https://doi.org/10.1007/978-3-031-72359-9_11
Download citation
DOI: https://doi.org/10.1007/978-3-031-72359-9_11
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-031-72358-2
Online ISBN: 978-3-031-72359-9
eBook Packages: Computer ScienceComputer Science (R0)