A Function Approximator Model for Robust Online Foot Angle Trajectory Prediction Using a Single IMU Sensor: Implication for Controlling Active Prosthetic Feet

Abstract

Lower limb dysfunction hinders the locomotive activity in individuals and compels them to use assistive devices such as prostheses or orthoses to restore the missing locomotive functions. Passive assistive devices have limitations in replacing the lost functionality. On the other hand, active devices could restore more natural locomotion using motorized joints. To make proper use of these embedded motors, a control architecture is required to translate the motion intent of the user into the intended prosthesis joint trajectories. Here, we propose a temporal convolution-based online foot angle trajectory prediction network (FATP-N) that predicts the foot angular positions during unconstrained walking from the shank angular position measured using a single wearable motion tracker sensor. The data acquisition experiments were designed to reflect natural gait with turns, varying inclines, speeds, and mixed cadences. The trained models were prepared for real-time prediction by compressing them to reduce the storage and time complexity. The validations were performed on different motion conditions, including those that have not been used for model training, to verify the robustness of the model. The proposed FATP-N achieved high accuracy and could generate predictions in real-time. The results indicate that the proposed approach offers a possible architecture for real-time control of powered intelligent prostheses.

RIOT: Recursive Inertial Odometry Transformer for Localisation from Low-Cost IMU Measurements

“…The data were captured via five different users and four different types of off-the-shelf consumer smartphones. The IMU data was collected and synchronised with a frequency of 100 Hz, which is generally accepted in various applications and research [ 48 , 49 , 50 ]. A high-precision optical motion capture system (Vicon) was used to capture full pose ground truth at m location and degree attitude accuracy [ 51 ].…”

A Deep Learning Model with a Self-Attention Mechanism for Leg Joint Angle Estimation across Varied Locomotion Modes

“…Consequently, this approach results in discontinuous motions during the transition stage between the locomotion tasks or gait phases [ 1 ]. In addition, the safety of device wearers can also be comprised by the latency and error in the classification of the activity or gait phase [ 12 ]. Irrespective of the above transition issues, these fixed pre-determined trajectories adapt poorly to real terrain variations that deviate from the structured walking environment found in laboratory settings [ 1 , 13 ].…”

“…Moreover, finite-state-based control strategies inevitably introduce many device parameters, thresholds, and switching rules to refine the division of the gait phases [ 12 , 14 ]. Au et al [ 15 ] conducted experiments to establish device settings and mode transition rules for different gait phase stages.…”

“…The advancement of deep learning enables a more generalized CLME-based trajectory planner to be created by learning non-linear motion patterns from daily motion data. Interest has been directed towards recurrent neural networks (RNN) [ 1 , 19 ], convolutional neural networks (CNN) [ 12 , 20 ], and dense neural networks (DNN) [ 21 ], as the main models for joint angle prediction. Instead of focusing on individual models for each activity, most studies aim to develop a comprehensive deep learning model that can deal with various activity modes.…”

“…Moreover, in contrast to a pre-defined trajectory, an appropriate gait pattern can be learnt for each user to adapt to their own stride length and cadence, which alleviates the issues related to inter-subject, intra-subject, and intra-gait variation [ 10 , 22 , 23 ]. Most of those approaches rely on inertial measurement unit (IMU) data to train the deep learning methods [ 10 , 12 , 20 , 22 , 23 ], while others aim to exploit the use of electromyography (EMG) [ 10 , 19 , 20 , 21 ]. The latter indicates that EMGs are more complex to use and better suited for classification tasks.…”