Bioformers: Embedding Transformers for Ultra-Low Power sEMG-based Gesture Recognition
Alessio Burrello1,a, Francesco Bianco Morghet2,c, Moritz Scherer3, Simone Benatti4, Luca Benini1,b, Enrico Macii5, Massimo Poncino2 and Daniele Jahier Pagliari2
1DEI, Università di Bologna, Bologna, Italy
aAlessio.Burrello@unibo.it
bLuca.Benini@unibo.it
2Department of Control and Computer Engineering, Politecnico di Torino, Turin, Italy
cFrancesco.Morghet@polito.it
dMassimo.Poncino@unibo.it
eDanieleJahier.Pagliari@polito.it
3Integrated Systems Laboratory, ETH Zurich, Switzerland
scheremo@iis.ethz.ch
4Department of Sciences and Methods for Engineering, University of Modena and Reggio Emilia, Italy
simone.benatti@unimore.it
5Interuniversity Department of Regional and Urban Studies and Planning, Politecnico di Torino, Turin, Italy
Enrico.Macii@polito.it
ABSTRACT
Human-machine interaction is gaining traction in rehabilitation tasks, such as controlling prosthetic hands or robotic arms. Gesture recognition exploiting surface electromyographic (sEMG) signals is one of the most promising approaches, given that sEMG signal acquisition is non-invasive and is directly related to muscle contraction. However, the analysis of these signals still presents many challenges since similar gestures result in similar muscle contractions. Thus the resulting signal shapes are almost identical, leading to low classification accuracy. To tackle this challenge, complex neural networks are employed, which require large memory footprints, consume relatively high energy and limit the maximum battery life of devices used for classification. This work addresses this problem with the introduction of the Bioformers. This new family of ultra-small attention-based architectures approaches state-of-the-art performance while reducing the number of parameters and operations of 4.9×. Additionally, by introducing a new inter-subjects pre-training, we improve the accuracy of our best Bioformer by 3.39%, matching state-of-theart accuracy without any additional inference cost.
Deploying our best performing Bioformer on a Parallel, Ultra- Low Power (PULP) microcontroller unit (MCU), the GreenWaves GAP8, we achieve an inference latency and energy of 2.72 ms and 0.14 mJ, respectively, 8.0× lower than the previous state-of-the-art neural network, while occupying just 94.2 kB of memory.
Keywords: Transformers, sEMG, Gesture Recognition, Deep Learning, Embedded Systems.
