LaViers steps up research on robotic choreography with extended DARPA award

06/13/2017
 
Amy LaViers
Amy LaViers
Following up on her 2016 Young Faculty Award from DARPA, Assistant Professor Amy LaViers was awarded extended funding for a third year that will continue to build on her research project, “Choreography of Embodied, Platform-invariant Motion Primitives.” 
 
The first two years of funded research for the project have focused on learning how embodied, qualitative techniques for choreography can influence the development of robotic control algorithms. LaViers and her team are creating a supervisory control system aligned with human movement. Their work, inspired by her expertise in Laban/Bartenieff Movement Studies (LBMS; a theory used in choreography), is centered on design of an embodied, platform-invariant notion of “movement” that will facilitate movement transfer between platforms. 
 
A novice user tester created movements that were then translated to two platforms: NAO  and Baxter.
A novice user tester created movements that were then translated to two platforms: NAO and Baxter.
Towards this goal, LaViers and her research group are developing a model for mapping spatial commands to physical platforms. Their unique method of producing complex motions using fewer commands has out-performed previous control methods, which require more, joint-specific parameters to be specified. 
 
Starting with embodied abstractions and collaborating with dance faculty and choreographers, they have developed notions of high-level motion patterns, inspired by LBMS Patterns of Body Organization, which allow different ways to move from one pose to another by changing robot configuration during movement. 
 
“This idea of formalism for expressivity will allow us to compare different platforms to each other and to animals (including humans) in a formal way. Our alternative approach – formulating robotic abstractions in a top-down way – is risky, but I believe it has great promise to give operators an unparalleled ease of control,” said LaViers.
 
During her additional year of funding, LaViers will perform further user studies in operational contexts to validate the kinesthetic empathy and efficiency of the architecture her team has developed. Utilizing virtual reality during this phase to assist more in-depth user testing, users will actively control the robots to perform simple tasks. Her research group is also continuing their training in LBMS. 
 

 

DARPA, the Defense Advanced Research Projects Agency, awards grants for research and technology that align with its mission of improving national security. The one-year Director’s Award extension LaViers received is offered to just a few YFA recipients each year. 
 
LaViers directs the Robotics, Automation, and Dance (RAD) Lab at the University of Illinois. She has worked in the area of advanced manufacturing through an industry-university consortium and has forged interdisciplinary ties with university dance programs and the Laban/Bartenieff Institute for Movement Studies, where she earned a certification in Laban Movement Studies.