BUILDING 3D-STRUCTURE WITH AN INTELLIGENT ROBOT SWARM

Abstract

This research is an extension to the TERMES system, a decentralized autonomous construction team composed of swarm robots building 2.5D structures1, with custom-designed bricks. The work in this thesis concerns 1) improved the mechanical design of the robots, 2) addition of heterogeneous building material, and 3) an extended algorithmic framework to use this material. In order to lower system cost and maintenance, the TERMES robot is redesigned for manufacturing in low-end 3D printers and the new drivetrain, including motor adapters and pulleys, is based on 3D printed components instead of machined aluminum. The work further extends the original system by enabling construction of 3D structures without added hardware complexity in the robots. To do this, we introduce a reusable, spring-loaded expandable brick which can be easily manufactured through one-step casting and which complies with the original robots and bricks. This thesis also introduces a decentralized construction algorithm that permits an arbitrary number of robots to build overhangs over convex cavities. To enable timely completion of large-scale structures, we also introduce a method by which to optimize the transition probabilities used by the robots to traverse the structure.