Sensor Fusion of Vision, Kinetics and Kinematics for Forward Prediction During Walking with a Transfemoral Prosthesis

We compared the performance of forward prediction systems for powered prostheses combining kinematics, kinetics, EMG, and vision while tested on a transfemoral amputee walking on a powered knee-ankle prosthesis.

In Review

Intent Prediction Based on Biomechanical Coordination of EMG and Vision-Filtered Gaze for End-Point Control of an Arm Prosthesis

We propose a novel controller for powered prosthetic arms, in which EMG and gaze data are used to predict the end-point for a full arm prosthesis, which could then be used to drive the forward motion of individual joints.

Link to TNSRE paper

Subject- and Environment-Based Sensor Variability for Wearable Lower-Limb Assistive Devices

We presented a device-agnostic, feature-driven approach and analysis of the variability within and between subjects and activities for each sensor modality for control and prediction of desired activity for powered prosthetic legs or exoskeletons.

Link to Sensors paper

Depth Sensing for Improved Control of Lower Limb Prostheses

We developed an algorithm to segment stairs from depth sensing of the environment and extract significant measures to detect the user intention, such as the distance and angle from the stairs, the number of steps, stair height, and stair depth. We conducted several experiments to test the our system and achieved 98.8% accuracy.

Link to TBME paper