Abstract:Locomotion transition recognition with external disturbances is a key issue in lower limb robotic prostheses. Redonning the prosthetic socket and individual differences are encountered frequently in daily life. They change the feature distribution and further fail the previously trained recognition model. To bridge the technical gap between laboratory validation and practical demands, researchers need to construct an adaptive recognition method with high accuracy, quick response, time-efficient calibration, and minimal interference to the human body. Our study developed an adaptive recognition algorithm based on two miniaturized inertial sensors (anterior thigh and forefoot) and a foot pressure sensor, which can easily be integrated into the prosthesis. The algorithm fused probability-based fuzzy classifiers, which took heuristic-based features as inputs with a dynamic-time-warping-based automatic template generation block. It enables the recognition model to quickly fit the parameters for an untrained mode after external disturbances. We validated the proposed adaptive recognition method on 13 healthy subjects, performing 18 motion transitions with random interday intervals and across user evaluation, both offline and online. Even with the data of one subject as the prior knowledge, the other subjects produced an average accuracy of 98.04% and an average transition time latency of -180.99 ms without manual training in interday and intersubject uses. We also tested the method on two subjects using robotic prostheses, one person with a transfemoral amputation and the other person with a transtibial amputation. The average accuracy was still as high as 98.06% (transfemoral) and 100% (transtibial) without manual calibration after seven days of intervals. The results proved that the adaptive recognition method is robust to sensor redonning and user changes. Compared with the state-of-the-art methods, our study makes a further step toward the practical use of robotic prostheses. Future studies will be conducted to implement the algorithm on the robotic prosthesis for more extensive tests.

Locomotion transition prediction at Anticipatory Locomotor Adjustment phase with SHAP feature selection

Ankle Foot Motion Recognition Based on Wireless Wearable Semg and Acceleration Sensors for Smart AFO

A Locomotion Intent Prediction System Based on Multi-Sensor Fusion.

Predicting Continuous Locomotion Modes via Multidimensional Feature Learning from sEMG

Taking Locomotion Mode As Prior: One Algorithm-Enabled Gait Events and Kinematics Prediction on Various Terrains

Locomotion Prediction for Lower Limb Prostheses in Complex Environments Via Semg and Inertial Sensors

Adaptive Locomotion Transition Recognition with Wearable Sensors for Lower Limb Robotic Prosthesis

Effective recognition of human lower limb jump locomotion phases based on multi-sensor information fusion and machine learning

Locomotion Mode Recognition with Robotic Transtibial Prosthesis in Inter-Session and Inter-Day Applications

Locomotion Mode Recognition with Inertial Signals for Hip Joint Exoskeleton

An enhanced ResNet deep learning method for multimodal signal-based locomotion intention recognition

A New Strategy for Parameter Optimization to Improve Phase-Dependent Locomotion Mode Recognition

Locomotion Mode Transitions: Tackling System- and User-Specific Variability in Lower-Limb Exoskeletons

On the Design of a Wearable Multi-sensor System for Recognizing Motion Modes and Sit-to-stand Transition

Locomotion Mode Recognition Based on Foot Posture and Ground Reaction Force

Motion Intention Prediction and Joint Trajectories Generation Toward Lower Limb Prostheses Using EMG and IMU Signals

A Foot-Wearable Interface for Locomotion Mode Recognition Based on Discrete Contact Force Distribution

Real-Time Onboard Human Motion Recognition Based On Inertial Measurement Units

Semg-Based Upper Limb Motion Recognition Using Improved Sparrow Search Algorithm

Motion Gait: Gait Recognition via Motion Excitation

Continuous Prediction of Lower-Limb Kinematics From Multi-Modal Biomedical Signals