Periodic Reporting for period 1 - MOrPhEM (Mechanics of Programmable Matter)
Período documentado: 2019-03-01 hasta 2021-02-28
In this project, the shape-shifters, also known as programmable matter, are viewed as structures composed of interconnected, microscopic, active robotic modules, which are able to process and exchange information, reconnect and move with respect to their neighbours. As such, they compose a computing network of continuously changing connection topology, which must collectively decide how to physically reorganise (similarly to fire-ants, which can form engineering structures from their bodies). One of the obstacles for them to freely reorganise is that, when shape-shifting proceeds, these physical computing collectives may experience a mechanical failure. Just like any other structure, a modular robot may break or turn over if it is not properly balanced. The goal of this project is to appropriately design and programme the collective to achieve both: structurally-safe and efficient transformations of its shape.
Although we physically demonstrated that the abovementioned scheme is capable of predicting mechanical failures for a variety of modular robotic structures, we also showed that the expected execution time of the algorithm grows extremely fast with the number of constituent robotic modules. Accordingly, with IPPT PAN, we developed an improved algorithm that partially reduces this efficiency bottleneck. This improved algorithm has been then validated on a simulator.
In order to transform shapes efficiently, a modular robot must form a special porous frame. This allows the remaining modules to reorganise in parallel across the robot’s volume (similar to water flowing through a foam). With FEMTO-ST, we are designing generic algorithms to guide that flow, which should be compatible with many existing- and possibly many future designs of modular robots.
The results of this project were disseminated in peer-reviewerd journals and conferences, as well as in the form of popular-science communications/articles.
One of beneficial effect of this project is the close collaboration that has been established between the teams from the University of Luxembourg, IPPT PAN and FEMTO-ST, as a part of the Programmable Matter consortium (https://www.programmable-matter.com/). This promises practical application of the results of this project and possible future commercialization.