Improving Robot Capabilities Through Reconfigurability

Abstract

Advancements in robot capabilities are often achieved through integrating more hardware components. These hardware additions often lead to systems with high power consumption, fragility, and difficulties in control and maintenance. However, is this approach the only path to enhancing robot functionality?
In this thesis, I introduce the PuzzleBots, a modular multi-robot system with passive mechanisms. Leveraging the inherent agility of individual locomotion, robots can collaborate to assemble into functional structures, reconfigure, and adapt to different environments. We show that we can enhance the physical capabilities of robot systems without significantly complicating the hardware design. We first utilize the environment’s structural features and forces. By using gravity as an activation force, we can implement passive mechanisms as connections between robots, without the need for additional power. By incorporating compliance within the robot assembly to improve traction, coupled robots can navigate challenging terrains more effectively. We then introduce our modular multi- robot systems, where the collective performance surpasses the capabilities of any single robot. By employing gravity as an activation force, we utilize passive mechanisms as connections between robots, without the need for additional power. Furthermore, we incorporate compliance within the robot assembly to improve traction, enabling coupled robots to navigate challenging terrains more effectively. We also utilize heterogeneity by combining different types of robots, where each one of them has its own strengths and weaknesses. Thirdly, we present our distributed model predictive control framework, which facilitates precise, real-time control over this highly constrained multi-robot system.
In summary, by utilizing the environment, coordinating an assembly of multiple robots, and controlling them efficiently, we can improve robot capabilities without complicating the hardware. We show the potential for simpler and more sustainable robot designs by showcasing the effectiveness of the PuzzleBot system, which uses fewer active components. I hope to encourage future works about the use of passive mechanisms and simple shapes to create efficient and functional robots.