*********************************
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
*********************************
Henry M. Clever
PhD Defense Presentation
Date: Friday, December 3rd, 2021
Time: 12:00 PM (EST)
Location: BlueJeans meeting (https://bluejeans.com/820441385/8474)
Committee:
Prof. Charlie Kemp (Advisor) – Department of Biomedical Engineering, Georgia Institute of Technology
Prof. James Hays – College of Computing, Georgia Institute of Technology
Dean Ayanna Howard – College of Engineering, Ohio State University
Prof. C. Karen Liu – Department of Computer Science, Stanford University
Prof. Greg Turk – Department of Interactive Computing, Georgia Institute of Technology
Title: Modeling Humans at Rest with Applications to Robot Assistance
Abstract: Humans spend a large part of their lives resting. Machine perception of this class of body poses would be beneficial to numerous applications, but it is complicated by line-of-sight occlusion from bedding. Pressure sensing mats are a promising alternative, but data is challenging to collect at scale. To overcome this, we use modern physics engines to simulate bodies resting on a soft bed with a pressure sensing mat. This method can efficiently generate data at scale for training deep neural networks. We present a deep model trained on this data that infers 3D human pose and body shape from a pressure image, and show that it transfers well to real world data. We also present a model that infers pose, shape and contact pressure from a depth image facing the person in bed, and it does so in the presence of blankets. This model similarly benefits from synthetic data, which is created by simulating blankets on the bodies in bed. We evaluate this model on real world data and compare it to an existing method that requires RGB, depth, thermal and pressure imagery in the input. Our model only requires an input depth image, yet it is 12% more accurate. Our methods are relevant to applications in healthcare, including patient acuity monitoring and pressure injury prevention. We demonstrate this work in the context of robotic caregiving assistance, by using it to control a robot to move to locations on a person’s body in bed.
