Assessing the effectiveness and improving the controllability of assistive robotic devices – from exoskeletons to prosthetic limbs – requires human movement data. Lots of it.
Some data collected in a lab might be from a person’s muscle signals that occur during contractions (EMG signals) or from the movement of a specific joint or limb (kinematics).
Unfortunately, there currently is no publicly available and unified database of bilateral kinematic and EMG data recorded from wearable sensors as people move freely (without an assistive robotic device) between different activities, like walking or climbing stairs. This type of data could help researchers compare different robotic devices, or improve the design of a device’s control system so that it can predict future states of movement (often referred to as “intent recognition”).
Blair Hu, a PhD student at the Center for Bionic Medicine, recently published a paper in Frontiers in Robotics and AI describing a dataset called ENcyclopedia of Able-bodied Bilateral Lower Limb Locomotor Signals (ENABL3S). The dataset contains EMG and joint and limb kinematics recorded from wearable sensors. Hu developed the dataset after analyzing data from 10 individuals as they completed different movement activities, ranging from sitting to walking and navigating ramps and stairs. The people tested did not have any gait impairments, but the experiments allowed Hu to collect important data about human movement.
“This dataset is important because it uses sensors commonly accessible to robotic devices to establish a baseline for how healthy individuals perform a wide range of walking activities. In the long-term, it will help us better evaluate how well a device is performing,” Hu said. “For example, if we’re testing a new robotic exoskeleton or orthosis designed to assist a person with muscle weakness, we can see how closely the person’s movement and muscle activity matches the information available in our dataset.”
Hu is mentored by Dr. Levi Hargrove at the Center for Bionic Medicine. His research focuses on using machine learning techniques to develop better control systems for robotic devices to improve mobility for individuals with lower-limb gait impairments.