Skip to main content

The Neuromechanics Lab focuses on restoring motor and sensory functions of the arm and hand impacted by a central or a peripheral injury, such as stroke and arm amputation. Specifically, we focus on four research topics in neural engineering and neural rehabilitation: Quantifying neuromuscular impairment of the arm & hand; Robust and continuous decoding of motor intention for dexterous control of assistive robots; Sensory encoding for human-in-the-loop control of assistive robots; and Neural stimulation to restore functional motor control of the hand.

Quantifying Neuromuscular Impairment of the Arm and Hand

After a central or peripheral injury, the neural and muscular systems are altered, which can lead to impairments (such as weakness, spasticity, and discoordination) in motor functions. An accurate quantification of these impairments over time is critical for prescribing rehabilitation or assistive strategies. We continuously monitor motor activities of the arm and hand using wearable sensing and signal processing techniques. We implement non-invasive high-density electromyographic (HD EMG) recording and processing techniques, in order to capture muscle activation patterns at the macro level (muscle or muscle groups) and at the micro level (individual motor units). We integrate neuromechanical information with standard clinical assessments to provide guidance for targeted and adaptive therapeutic and assistive strategies.

An illustration of the spatial activation patterns of the finger extensor digitorum muscles, and the spatial distribution of individual motor unit action potentials with corresponding energy distribution

Robust Decoding of Motor Intention for Dexterous Control of Assistive Robots

Highly functional rehabilitative and assistive robots are commercially available. However, the clinical utility of these advanced devices is limited, largely due to a lack of robust and intuitive neural-machine interface. We perform continuous neural decoding of the motor command sent from the brain to the muscles, using various signal processing and machine learning techniques. The decoder performance is robust to environmental interference and muscle signal variations. The decoded motor commands, either for fingertip force or joint kinematic control, can drive advanced robotic hands to enable dexterous functional movements.

Neural decoding based on motor unit decomposition and probability of populational neuron firing frequency

Real-time decoding to predict fingertip forces with robust performance over time, when compared with conventional EMG-amplitude based approach

Sensory Encoding for Human-in-the-Loop Control of Assistive Robots

An effective use of our hand involves different levels of bi-directional sensory-motor interactions. A majority of prosthetic hand users, however, receive limited somatosensory feedback when they interact with their prostheses. A lack of somatosensory feedback can limit functional utility of advanced prosthetic hands. To address this longstanding issue, we develop and evaluate sensory stimulation techniques to close this sensory loop. We develop a peripheral nerve stimulation technique to provide intuitive haptic information of the phantom hand to prosthetic hand users, and we implement vibrotactile stimulation to provide proprioceptive information of joint kinematics to prosthetic hand users. The evoked somatosensory feedback improves object manipulation and recognition of object properties.

Selective haptic sensations elicited in the phantom hand through electrical stimulation of the sensory nerves. The shaded regions represent the location of the sensation evoked from different electrodes

Object size and stiffness recognition performed by an individual with an arm amputation, using concurrent fingertip force feedback through nerve stimulation and joint proprioceptive feedback through vibrotactile stimulation

Neural Stimulation to Restore Functional Motor Control of the Hand

Hand dexterity allows us to perform a variety of tasks in our daily activities. However, a loss of such critical skill is common in a majority of individuals with neurological conditions, and existing efforts to restore such function have limited success. We develop non-invasive neural stimulation techniques to activate neural tissues (peripheral nerve or spinal cord), in order to restore motor control of the hand. We use spatially and temporally organized electrical stimulation patterns to activate a selective number of nerve fibers in a biomimetic manner. The stimulation approach can elicit dexterous finger and wrist movements, and enable daily hand use in individuals with neurological disorders.

Elicited various hand grasp patterns through median and ulnar nerve stimulation


Elicited finger and wrist extension through radial nerve stimulation