FURI | Spring 2024

Mesh Based Machine Learning Algorithms for Use in Deforming Solids Simulations

The goal of this research is to determine the effectiveness of graph neural networks in predicting the deformation and motion of solids. For this research specifically a two-dimensional beam was used, however, the long-term goal is to apply these methods to three-dimensional soft robot models. The graphical neural network model used functions on a system of nodes and edges that make up the physical objects mesh. By training the model using data from traditional physical simulations it is predicted that the trained model will be able to correctly predict the deformation of the beams with various elasticities.