A Comparison of Pattern Recognition Control and Direct Control of a Multiple Degree-of-Freedom Transradial Prosthesis
With existing conventional prosthesis control (direct control), individuals with a transradial amputation use two opposing muscle groups to control each prosthesis motor. As component complexity increases, subjects must switch the prosthesis into different modes to control each component in sequence. Pattern recognition control offers the ability to control multiple movements in a seamless manner without switching.
Evaluation of linear regression simultaneous myoelectric control using intramuscular EMG
We demonstrated the ability for linear regression-based methods to decode patterns of muscle co-activation from intramuscular electromyography and provide simultaneous control of a wrist and hand system for future application in advanced powered prosthetic limbs. Real-time controllability was evaluated using able-bodied subjects participating in a virtual target-acquisition task, where intended wrist and hand motion corresponded to the movement of a ring-shaped cursor.
Gait Characteristics When Walking on Different Slippery Walkways
The ability to change gait patterns in the presence of a slippery surface is essential for minimizing the risk of a slip and fall. By characterizing changes in lower-limb muscle activity and kinematics of the able-bodied population we can gain an initial estimate of how a prosthetic limb should behave on slippery surfaces to minimize the users risk of slipping.
Depth Sensing for Improved Control of Lower Limb Prostheses
We developed, characterized, and validated an algorithm for recognizing stairs in the environment using data from a worn RGB-D sensor. The measures that we extracted from the environment, including the distance to the stairs, angle of approach, height, width, and depth of stairs, and stair count, were characterized and found to be highly correlated and accurate. Also, an estimate of when the user was approaching stairs was produced during an online walking test, which resulted in over 98% accuracy and a frame rate of more than 5 fps. We plan to fuse the environmental estimates with information obtained from EMG, kinetics, and kinematics for predicting the correct locomotion mode for an ankle-knee prosthesis.