Prediction Algorithm of Parameters of Toe Clearance in the Swing Phase

Area of Science:

• Robotics
• Biomechanics
• Machine Learning

Background:

• Adaptive control in gait training robots aims to enhance performance by assisting motion.
• Predicting toe motion is crucial for mitigating tripping risks, a challenge not addressed by conventional robotics.
• Toe clearance prediction during walking is key to reducing the likelihood of trips.

Purpose of the Study:

• To propose a novel method for predicting toe clearance during walking using a radial basis function network.
• To enhance the safety and effectiveness of adaptive gait training robots.
• To improve the accuracy of predicting both maximum and minimum toe clearances within a single swing phase.

Main Methods:

• Utilized a radial basis function network for toe clearance prediction.
• Input data included hip, knee, and ankle joint angles, angular velocities, and accelerations in the sagittal plane at the swing phase onset.
• Trained and tested the network with gait data from multiple subjects walking on a treadmill.

Main Results:

• Achieved a root mean square error of 3.28 mm for maximum toe clearance and 2.30 mm for minimum toe clearance.
• Demonstrated high prediction accuracy even when walking velocity changed, with errors of 4.04 mm (max) and 2.88 mm (min).
• Outperformed existing methods in prediction accuracy for toe clearance.

Conclusions:

• The proposed algorithm accurately predicts future maximum and minimum toe clearances within the same swing phase.
• Leveraging joint movement information at the start of the swing phase enables precise toe clearance prediction.
• This method offers a significant advancement for adaptive robotic gait training systems, enhancing safety and efficacy.