High-level Perception and Control for Autonomous Reconfiguration of Modular Robots (NSF CPS)

PI:
Hadas Kress-Gazit
Co-PI:
Mark Campbell, Mark Yim (UPenn)
Funding:
NSF CNS-1329692
Active Dates:
2013-2016
People:
Goal:
The goal of this research is to develop theory, hardware and computational infrastructure to enable automatically transforming user-defined, high-level tasks into correct, low-level perception informed control and configurations for modular robots.
Tasks:
1. Hardware: designing and building a robust reconfigurable hardware platform based on UPenn's SMORES work.
2. The Library: develop (manually and automated tools) perception, locomotion, manipulation and reconfiguration components as well as a mapping between the low-level control of a modular robot and basic properties (semantics and continuous metrics).
3. High-level control synthesis: Develop methods for synthesizing high level, information informed, controllers for modular robots based on the library entries, their correct sensor based sequencing, and a grammar and parser designed to address the unique characteristics of modular self-reconfigurable robots.