Super Cards
We are pleased to announce that five LARRI students won for their poster presentations at this year’s KY NSF EPSCoR Super Collider event, organized and run virtually by University of Kentucky on February 26, 2021. ECE Undergraduate Brooke Ritz, ECE Master’s students Cameron O’ Nan, Nathan Hodges, and Jacob Berdichevsky, along with BE Doctoral student Johnathan George were all announced winners in their perspective categories.
|
Cameron O’ Nan
|
Nathan Hodges
|
Jacob Berdichevsky
|
Johnathan George
|
Brooke Ritz
Cameron O’ Nan’s presentation outlined his research in improving ARNA’s environmental awareness. His independent study involved analyzing the improvements that could be made to the ARNA sensor suite and designing improvements to allow ARNA to operate in autonomous modes more confidently. To achieve this goal, he had to implement and verify higher quality data channels on the ARNA robot. In order to make ARNA environmentally aware, Cameron handled the development and measurement of the sensor channel capacity. He identified measurable data points to compare results of packet loss overtime, max sampling frequency, latency, emulate real driving environment and analyze results of the new board. His new design gave the robot higher reliability data, faster communication, and an increased safety factor than previous designs.
Nathan’s poster won for his investigation into future robot skin technology for pHRI. Nathan’s focus is on Force Sensitive Resistors (FSR) and Metal Foil Strain Gauges (MFSG) and how applicable they could be to pHRI together and individually. His research will include determining how useful FSR technology would be for accurately and reliably reading and locating externally applied forces. In addition, he would like to investigate whether fusion of FSR tech with metal foil strain gauge technology could provide any practical applications for pHRI. Nathan’s experiment will include testing commercial FSR/MFSG individually and together in similar environments, testing the sensor spatial resolutions response when mounted to various material stiffness, and the response when covered with varying material stiffness. The goals detailed in the presentation are a comprehensive summary of fundamental differences in FSR/SG performance and discover which tech is better suited for a particular interface/application.
Jacob’s poster details a novel approach based on deep reinforcement learning for robot adaptive motion control. Collaborative robot systems are important in facilitating human robot interaction in an industrial environment. This research proposes a novel online algorithm for motion similarity measurements during human-robot interaction (HRI). Using Deep Deterministic Policy Gradients (DDPG) and Dynamic Movement Primitives (DMP) will optimize the robot’s ability to teach specific motions. With the DMP a library of teachable motions can be created. A reward function based on Dynamic Time Warping (DTW) will be used in DDPG to enhance human robot interaction. The third part of the research thrust is to focus on the development of new control techniques for human in the loop robotic systems.
Johnathan is currently working with children who have spinal cord injuries (SCI), and using a custom-designed rocking chair equipped with integrated sensors to learn about how their muscles are activating while they rock in the chair. The goals of this preliminary study are to verify muscle activation in the arms, legs, and trunk of the children while rocking, compare electro-myography (EMG) data to rocking chair sensor data, and obtain feedback from the children, parents, and therapist on the prototype design of the rocking chair. The objective of this research is to improve and monitor trunk control of children with SCI. Johnathan says, “Ultimately, we’d like to use the sensors on the rocking chair not only to detect the child’s muscle activation, but also to feed back into actuators which could adjust the force required to rock the chair to match the child’s level of ability.”