Human Gait Data Augmentation and Trajectory Prediction for Lower-Limb Rehabilitation Robot Control Using GANs and Attention Mechanism

Abstract

To date, several alterations in the gait pattern can be treated through rehabilitative approaches and robot assisted therapy (RAT). Gait data and gait trajectories are essential in specific exoskeleton control strategies. Nevertheless, the scarcity of human gait data due to the high cost of data collection or privacy concerns can hinder the performance of controllers or models. This paper thus first creates a GANs-based (Generative Adversarial Networks) data augmentation method to generate synthetic human gait data while still retaining the dynamics of the real gait data. Then, both the real collected and the synthesized gait data are fed to our constructed two-stage attention model for gait trajectories prediction. The real human gait data are collected with the five healthy subjects recruited from an optical motion capture platform. Experimental results indicate that the created GANs-based data augmentation model can synthesize realistic-looking multi-dimensional human gait data. Also, the two-stage attention model performs better compared with the LSTM model; the attention mechanism shows a higher capacity of learning dependencies between the historical gait data to accurately predict the current values of the hip joint angles and knee joint angles in the gait trajectory. The predicted gait trajectories depending on the historical gait data can be further used for gait trajectory tracking strategies.

Force Transmission Analysis and Optimization of Bowden Cable on Body in a Flexible Exoskeleton

“…Furthermore, the simulation result showed that the optimum efficiency was 78.68% when the length of Bowden cable between the two sheath brackets was 475 mm (4) A flexible exoskeleton prototype was assembled to evaluate the theoretical analysis, and the experiment results suggested that the force transmission efficiency had a strong relationship to the length or curve angle of the Bowden cable, instead of the transmission velocity. The optimal efficiency and length of Bowden cable from this prototype experiment showed consistency with the modeling and simulation in this paper (5) The metabolic cost test not only presented the reduction of human consumption with the knee assistance but also indicated that the effectiveness of our exoskeleton and its optimization of the Bowden cable…”

“…Exoskeleton is a kind of intelligent wearable robot that can assist the upper or lower limbs of human movement. With the development of its technology, exoskeletons have been widely used in medical rehabilitation, logistics, and military fields [1][2][3][4][5]. Based on the structure mode, exoskeletons are divided into rigid assist [6][7][8] and flexible assist [9][10][11][12][13][14].…”

Deep learning with an attention mechanism for continuous biomechanical motion estimation across varied activities

“…Owing to the rare application of the attention mechanism method in the domain of gait kinematics estimation, many researchers cite works from other fields such as NLP or stock price forecasting. Rai et al (2020) and Wang et al (2021) adopted an autoregressive attention model developed for NASDAQ 100 Index forecasting ( Qin et al, 2017 ), whereby the calculation of attention score is equivalent to the encoder-decoder attention model used for machine translation in NLP. Specifically, the current hidden state in the decoder queries the hidden state sequence in the encoder and judges its importance when the decoder is outputting a sequence of results ( Bahdanau et al, 2014 ).…”

“…Specifically, the current hidden state in the decoder queries the hidden state sequence in the encoder and judges its importance when the decoder is outputting a sequence of results ( Bahdanau et al, 2014 ). However, Rai et al (2020) and Wang et al (2021) used the motion history of target limb as model inputs to make future prediction for the same target limb, which is not applicable for amputees. Li et al (2021) and Zhu et al (2021) weighted input features through adding an attention layer at the beginning before passing them into the LSTM layer, but do not provide a detailed description of the implementation of their feature attention mechanism.…”

Design and Control of Advanced Mechatronics Systems

“…The next four papers focused on the research on rehabilitation robots. In [8], a GANsbased (Generative Adversarial Networks) data augmentation method was created to generate synthetic human gait data while still retaining the dynamics of the real gait data. Then, both the real collected and the synthesized gait data were fed to our constructed two-stage attention model for gait trajectories prediction.…”