Data-driven methods based on neural networks are becoming more widespread for predicting human lower-limb motion. Until now, however, actual examples have focused on only a handful, steady locomotion behaviors. Here we explore if neural network predictors can simultaneously cover many more behaviors including transient ones. Training four common types of predictor networks on a large data set of human gait, we find that they all accommodate these behaviors similarly well, maintaining prediction errors of a few centimeters (lower-limb joint positions) and degrees (joint angles) when tested on data of previously seen subjects. We further observe that although the prediction quality drops for data of unseen subjects, overall, the predicted and actual lower-limb motions remain well aligned. While the predictors demonstrated here cover the largest range of locomotion behaviors reported to date, we achieve this improvement not by better network design but simply by training on more data. This outcome clearly supports the notion that the fastest route to obtain truly general network predictors of lower-limb motion is by focusing time and effort on the rapid growth and sharing of data sets of locomotion behaviors encountered in daily life.

We are developingg a novel learning framework for wearable robots that prefer task-invariant, energetic control rather than task-specific, kinematic control. By shaping the total energy of the human body with exoskeleton actuators, reducing mass/inertia parameters in body energetics to dynamically offload the weight of a patient. This framework utilizes optimization methods like CMA-ES and BO to identify "optimal" control parameters for task-invariant tasks. Using powered orthoses to assist humans in a variety of activities is coming true.

Our work aims to increase the stability of the DS gait by investigating the use of an inertial tail-like appendage. We model the quadruped system as a self-manipulator with an actuated tail in 3D which can swing in not only the yaw direction but also the pitch direction.

Our work investigates the use of a heavy tail to help stabilize larger-stride-displacement gaits for a quadruped robot. With the help of an actuated tail, such a quadruped can take advantage of the higher stride displacement (about 30% larger) of the diagonal sequence walking gait while maintaining static stability.

We use geometric mechanics to prescribe gaits with the coordination of leg movements and back-bending motion. We are able to show back-bending in quadruped, especially in salamanders, can improve stride displacement in the forward, rotational and lateral directions. Also, by assuming all stance feet having zero linear velocities, we model the quadruped locomotion on flat ground. We perform the simulation and robot test to analyze how back-bending improves locomotion.