tiSPHi: A parallelised GPU-accelerated SPH framework for modelling robot-ground interaction
tiSPHi: A parallelised GPU-accelerated SPH framework for modelling robot-ground interaction
Efficient locomotion of robots in real environments involves highly complex robot-ground interaction (RGI) problems. Granular geomaterials exhibit complex multiphase phenomena under dynamic impact, such as solid-like and fluid-like characteristics, together with an ejected gas-like phase. These phenomena occur in diverse situations, ranging from impact and penetration problems to locomotion over natural or man-made terrains. This range of behaviours is yet difficult to simulate accurately and efficiently. Here, we develop tiSPHi, a continuum model based on smoothed particle hydrodynamics (SPH) using Taichi parallel computing language. Using a simple Drucker-Prager model of geomaterials, we demonstrate a computationally efficient rapid modelling of granular locomotion of arbitrary intruders. Several tests are carried out to examine the capability of this model. Results show that this model provides a good computational efficiency with a balance of development productivity compared to other implementations, whilst, outperforming in terms of accuracy.
Zhibin Lei; Raul Fuentes
10th European Conference on Numerical Methods in Geotechnical Engineering (NUMGE2023)
2. Finite element, finite difference, discrete element, material point and other methods
https://doi.org/10.53243/NUMGE2023-279
Contact Information
- Secretary General - Dr A M McNamara
- Email: [email protected]
- Phone: +44 20 7040 8154
- City, University of London
- Northampton Square
- London EC1V 0HB. United Kingdom