Publications

Current therapy tools have limited clinical outcomes for upper-limb recovery due to incomplete models of hand and wrist function that in turn inform device and assessment designs. The purpose of this thesis is to investigate the design of devices and protocols that leverage robotic abilities to collect large datasets for the testing of open hypotheses in neuroscience and functional perception. Starting with finger mobility, I introduce novel miniature sensors for a tabletop device that records human fingertip forces in 3D task space and in real-time. The goal of the sensors and the larger device is to track and identify abnormal manipulation that may occur in a human hand post-stroke. A rich characterization of the device’s sensing capabilities is described, and its implications for dexterity research in identifying finger deficits and treating them through targeted training are discussed. Then I focus on wrist mobility in the context of teleoperation. Specifically, I summarize the field of teleoperation research, and I focus on findings regarding tool compensation and human tracking. To understand the limits of wrist tracking during teleoperation, I collect performance data from healthy controls attempting to track multiple speed profiles through different dynamics on an existing teleoperator testbed. Our team found that users adjust their wrist control strategy depending on device dynamics, an ability that breaks down as tracking speed gets more difficult. Therefore, there are quantifiable limits to rotational compensation at the wrist. Lastly for wrist mobility, I discuss how aspects of human proprioception can dissociate depending on context, and I present the concept of a sensory “fingerprints” as a framework for identifying patient-specific patterns of deficit across these aspects. I introduce a new device for testing wrist proprioception and functional ability, and I conduct an assay of multiple psychometric tests and task simulations on healthy controls to find associations between aspects. Our team only found one correlation in rank between passive velocity discrimination and active position matching, and we found that passive and active velocity senses do differ. These findings support the assertion that our device is capable of collecting data needed to generate sensory “fingerprints” for targeted treatment.

https://jscholarship.library.jhu.edu/handle/1774.2/69157