Reconfiguration algorithms for robotically manipulatable structures

摘要

This paper addresses the design and optimization of robotically-reconfigurable structures from an algorithmic point of view. First, we address the algorithmic challenge of searching for a sequence of structural modifications that can transform a given modular structure into a new target structure that serves a different function. The target structure is not explicitly specified, only its desired function, and so the planning algorithm needs to account for both design and the corresponding deconstruction and construction sequence simultaneously. A number of different internal representations are considered and compared. We then demonstrate the algorithm under uncertainty in resource availability, where components may become missing during the build. Finally, we physically implement a robotically manipulatable structure design and manually confirm the reconfiguration process. We suggest that a combination of reconfigurable structures, robust reconfiguration algorithms that function under resource uncertainty, and reconfiguration robots, could open the door to a metabolic process where structures are decomposed and recomposed autonomously to meet varying needs for a variety of applications from infrastructure recovery to space exploration.