PhD student Jaemin Lee defended his PhD Thesis addressing problems at the interesction of trajectory generation, optimization and control of humanoid robots. You can find a link to his PhD thesis here:
https://www.dropbox.com/s/t5gkapun1rkxtni/2022-05-05-FINAL-Dissertation_Jaemin.pdf?dl=0
Our paper “Mixtures of Controlled Gaussian Processes for Dynamical Modeling of Deformable Objects” has been accepted for the Learning for Dynamics & Control Conference, 2022. Congratulations to Ce Xu and our collaborators at the Polytechnic University of Catalonia (UPC)!
Here is a video of the content of the class and selected project presentations:
Ye Zhao, Yinan Li, Luis Sentis, Ufuk Topcu, Jun Liu, Reactive Task and Motion Planning for Robust Whole-Body Dynamic Locomotion in Constrained Environments, The International Journal of Robotics Research, In Press 2022
This study takes a first step toward formally and reactively task planning and whole-body dynamic loco-manipulation behaviors in constrained and dynamically changing environments. We formulate a two-player temporal logic game between the multi-limb locomotion planner and its dynamic environment to synthesize a winning strategy that delivers symbolic locomotion actions. A controller further executes low-level motion primitives that generate feasible locomotion trajectories.
In this video, three ground robots — two A1 quadrupeds and one HSR mobile platform — sweep an environment quickly in search of a missing volleyball ball. They achieve this capability by employing online path planning and heterogeneous clustering to accommodate for the different speeds and fields of view of each robot. The algorithm supports up to 50 robot/vehicle search and is robust to robot/vehicle failures by re-planning at runtime. This technology is geared towards tasks such as emergency response where people could be missing during a fire or flood.
This work describes an online gain adaptation method to enhance the robustness of whole-body controllers for legged robots under unknown external force disturbances. Without properly accounting for external forces, the closed-loop control system incorporating WBC can easily become unstable, and therefore the desired task goals may not be achievable. The proposed method serves as a low-level controller for tracking whole-body trajectories more robustly than using fixed gain feedback control methods. Link to abstract:
https://www.frontiersin.org/articles/10.3389/frobt.2021.788902/abstract
Our lab is announcing multiple NSF Ph.D. Fellowships in Ethical AI starting 2022 through 2025! Good Systems, a UT Grand Challenge, and Texas Robotics are recruiting our first cohort of NSF Research Traineeship Ph.D. Fellows in Ethical AI. Prospective students interested in working in the HCRL can apply for doctoral admission in our department. More information can be found under the Fellowship tab: http://shorturl.at/gAEY9.
In this video, PI Luis Sentis talks about efficient actuation approaches for human-centered robots:
Congratulations to Ce Xu, our recent student co-advised with the UPC for devising new methods to learn and predict the motion of soft materials such as towels using mixture of Gaussian Processes. More details will be available soon.
The HCRL is proud to share our new NSF National Research Traineeship on Ethical AI Training for Roboticists. Congratulations to all the participants and to the Good Systems Bridging Barriers Program.
