*********************************
There is now a CONTENT FREEZE for Mercury while we switch to a new platform. It began on Friday, March 10 at 6pm and will end on Wednesday, March 15 at noon. No new content can be created during this time, but all material in the system as of the beginning of the freeze will be migrated to the new platform, including users and groups. Functionally the new site is identical to the old one. webteam@gatech.edu
*********************************
Title: Ultrasonic Imaging and Tactile Sensing for Robotic Systems
Committee:
Dr. David Taylor, ECE, Chair, Advisor
Dr. Gregory Durgin, ECE
Dr. Yorai Wardi, ECE
Dr. Ying Zhang, ECE
Dr. Aldo Ferri, ME
Abstract: This research develops several novel algorithms that enhance the operation of ultrasonic and tactile sensors for robotic applications. The emphasis is on reducing the overall cost, system complexity, and enabling operation on resource-constrained embedded devices with the main focus on ultrasonics. The research improves key performance characteristics of pulse-echo sensor systems – the minimum range, range resolution, and multi-object localization. The former two aspects are improved through the application of model-based and model-free techniques. Time optimal principles precisely control the oscillations of transmitting and receiving ultrasonic transducers, influencing the shape of the pressure waves. The model-free approach develops simple learning procedures to manipulate transducer oscillations, resulting in algorithms that are insensitive to parameter variations. Multi-object localization is achieved through phased array techniques that determine the positions of reflectors in 3-D space using a receiver array consisting of a small number of elements. The array design and the processing algorithm allow simultaneous determination of the reflectors, achieving high sensor throughputs. Tactile sensing is a minor focus of this research that leverages machine learning in combination with an exploratory procedure to estimate the unknown stiffness of a grasped object. Gripper mechanisms with full-actuation and under-actuation are studied, and the object stiffness is estimated using regression. Sensor measurements use actuator position and current as the inputs. Regressor design, dataset generation, and the estimation performance under nonlinear effects, such as dry friction, parameter variations, and under-actuated transmission mechanisms are addressed.
