2019 Position-based Ice Melting (Work in Progress)
In this project, we incorporate Position-based Dynamics for simulating particle-particle interaction with phase transition. The code is still buggy as it appears some completely melted ice particles still yield water particles dropping out of the invisible portion of the bunny's right ear.
2019 CS Undergrade Projects: Self-Driving Car using Imitation Deep Learning
Thamolwan Poopradubsil and Phornrawin Chitsoonthorn
Project Advisor: Chakrit Watcharopas, Co-Advisor: Usa Sammapun and Ruj Akavipat
Self-Driving Car using Imitation Deep Learning is a senior research project conducted by Thamolwan and Phornrawin under my supervision. At the current stage of the project as shown in the video, a trained vehicle is able to drive in the lane, make turns, and follow the traffic lights.
2018 Ice Melting Simulation using SPH and Heat Transfer with Constant Ambient Temperature
Nutcha Chayanurak and Chakrit Watcharopas
Abstract. In this paper we propose a particle-based approach to ice melting simulation using heat received from the air assuming to be around an ice object using Newton’s law of cooling. We use the SPH method for handling ﬂowing motion of water coming off from the melting ice surface. However, for melted water that is sparsely generated and ﬂowing on the ice surface, thin features are difﬁcult to simulate only with the SPH-based approach. To avoid unnatural appearance of a few water particles ﬂowing on ice surface, we extract an isosurface from the density distribution of the desired characteristic calculated from melted ice volume transferring to water volume, based on the ratio between current heat of the ice and latent heat of the ice fusion. We also explain a simplicity of our simulation method that reduces computational cost during the heat transfer process.
2015 Extracting Surface Geometry from Particle-Based Fracture Simulations
Chakrit Watcharopas, Yash Sapra, Robert Geist, and Joshua A. Levine
Abstract. This paper describes an algorithm for fracture surface extraction from particle-based simulations of brittle fracture. We rely on a tetrahedral mesh of the rest con guration particles and use a simple, table-lookup approach to produce triangulated fracture geometry for each rest con guration tetrahedron based on its con guration of broken edges. Subsequently, these triangle vertices are transformed with a per particle transformation to obtain a fracture surface in world space that has minimal deformation and also preserves temporal coherence. The results show that our approach is e ective at producing realistic fractures, and capable of extracting fracture surfaces from the complex simulation.
The video shows an experiment of a solid marble sphere flying into a concrete Stanford Armadillo.
This second video shows a projectile being shot through a glass plate.
The video shows the effectiveness of per particle transformation over per component transformation. The left vase is tranformed using per component transformation while the right is tranformed with per particle transformation.
The fourth video shows a hand smashing a stack of plates, demonstrating a complex simulation of multiple object interactions. The fracture patterns occurred inside each plate in rest position are also shown on the top right corner.
Visualizing 6 simulations simultaneously.
Fracture Propagation Visualization
Visualizing Connected Components
Visualizing Particle Velocity