A method for fluid-guided rigid body control based on deep reinforcement learning

A reinforcement learning and rigid body technology, applied in the fields of computer graphics fluid simulation, reinforcement learning, and optimal control, it can solve problems such as inability to be physically and accurately compatible with fluid simulators, inability to control dynamic rigid bodies, and difficulty in completing control tasks.

Active Publication Date: 2021-03-26
NANKAI UNIV
View PDF5 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0005] The purpose of the present invention is to solve various problems existing in the existing fluid control methods, including the inability to be physically and accurately compatible with all fluid simulators, the inability to control the dynamic rigid body in the fluid-rigid body coupling simulation area, and the difficulty in completing relatively complex control tasks. A method for fluid-guided rigid body control based on deep reinforcement learning

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • A method for fluid-guided rigid body control based on deep reinforcement learning
  • A method for fluid-guided rigid body control based on deep reinforcement learning
  • A method for fluid-guided rigid body control based on deep reinforcement learning

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0043] In this embodiment, the scene where 2-dimensional fluid and solid coexist is used, and the fluid nozzle is controlled to add fluid at the boundary according to the deep reinforcement learning algorithm. Finally, the effect of balancing the solid in the center of the screen is achieved through a control method based on physical laws. This scene produces lively and interesting animation effects. The overall implementation framework is as figure 1 As shown, firstly, the rigid body features are extracted from the state of the physical simulator and the encoded fluid features after dimensionality reduction are input into the strategy MLP network to output control actions, which are supplied to the simulator to execute actions and generate the next state. Among them, the TRPO algorithm is used to optimize the policy network, and the network structure is as follows: image 3 shown. In this embodiment, the scene and key parameters such as Figure 4 as shown, Figure 5-Figur...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

No PUM Login to view more

Abstract

The invention discloses a rigid body control method for fluid guide based on deep reinforcement learning. In the method, the behavior of a fluid-rigid body simulator is changed by applying a control force only at the boundary of a simulated region, and the fluid is controlled by a Navier-Stokes equation and the rigid body is controlled by a Newton-Euler equation in the simulated region. The controller of the method is a neural network trained by deep reinforcement learning, and through pre-training, a control action can be generated in an online mode. The controller based on the method receives states of the fluid and the rigid body as input, a fluid nozzle is controlled to move at the boundary and injects fluid towards the rigid body in the simulated region, and thus, physical realistic simulation results can be generated, and good effects are obtained in multiple two-dimensional fluid-rigid control tasks. The method can also expanded to a three-dimensional fluid-rigid coupling system, for example, the rigid body can be controlled to accurately move to a specified three-dimensional target point.

Description

technical field [0001] The invention relates to the technical fields of computer graphics fluid simulation, optimal control and reinforcement learning, in particular to a method for controlling a fluid-rigid body coupling system. Background technique [0002] Fluid simulation has been extensively studied in the field of computer graphics and there have been many different technical approaches. Fluid simulators can be categorized according to discretization methods, including grid-based methods used in "Animation and Rendering of Complex Water Surfaces" by Enright et al., 2002, and "Weakly Compressible SPH for Free Surface Flows" by Becker and Teschner, 2007. The particle-based approach used in the paper, and the hybrid approach used in the paper "Two-Way Coupled SPH and Particle Level SetFluid Simulation" by Losasso et al., 2008. Fluid simulators can also be classified according to the underlying integration method, including the explicit integration method used in "Weakly ...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Application Information

Patent Timeline
no application Login to view more
Patent Type & Authority Patents(China)
IPC IPC(8): G05B13/04
CPCG05B13/042
Inventor 任博田韵声马平川潘哲融
Owner NANKAI UNIV
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap