LogoLogo
ISTVSConferences
  • 16th European-African Regional Conference of the ISTVS
  • program
    • Overview
    • Preconference Professional Short Course
      • Elements of Planetary Rover Design
    • Student projects
    • Lublin Young Volunteers Program
  • Conference
    • Call for Papers
    • Registration
    • Location
    • Technical tour
    • Statement on Publication Ethics and Malpractice
    • Contact
  • Submissions
    • Papers
      • 0274 / Deep learning method for IMU-based tracking of Martian rover
      • 0448 / Lunar rover discrete element method study and calibration
      • 0709 / Proposal of swarm rovers’ collaborative locomotion with expansion and contraction...
      • 0889 / Skidding suppression method using “discrete 4-wheel-drive typed rover” considering...
      • 1098 / Simulation of change in supporting force when imparting vibration by distinct element method
      • 1453 / Suppressing the reduction of the traveling displacement on loose soil for rovers with...
      • 1522 / Tractive performance of rigid wheel in granular media using coarse-scale DEM models
      • 1638 / Analysis of tire characteristics on road surface with volcanic ash fall
      • 1805 / Accelerating graph networks for real-time physics simulations
      • 2145 / Obstacle performance and wheel failure test analysis of Zhurong rover
      • 2177 / Effects of road wheel load, driving speed and track slip upon stress state in sandy...
      • 2437 / Improving predictive control methods on off-road vehicles with realistic steering preview...
      • 2765 / Obstacle detection vision system enabling autonomous mounding on clearcuts
      • 2801 / In Situ Soil Property Estimation for Autonomous Earthmoving Using Physics- Infused Neural...
      • 2936 / Moisture content impacts on soil load bearing capacity and its spectral behaviour
      • 3280 / Comparison of tweels and pneumatic tires on LTATV military vehicle
      • 3493 / Numerical modeling of a tire on undrained saturated clay using FEM, ALE, and SPH
      • 3507 / Determination of the tractive forces of a tractor based on the reconstruction of...
      • 3796 / The issue of assessing the suitability of own simulation models for testing off-road vehicles
      • 3797 / Disturbed and naturally recovered soil surface as a ground for subsequent vehicle mobility...
      • 3835 / Effect of steer angle rate upon tyre lateral force generation on two different soils
      • 4141 / Modeling and verification of a full-scale forestry vehicle real-time multi-physics digital tw
      • 4456 / Modeling soil-tool interaction of a cultivator sweep using DEM
      • 4595 / Evaluation and comparison of driving performance of a lunar exploration rover wheel in...
      • 4681 / Simplified models of terrain-vehicle interaction for real-time applications
      • 4816 / Effects of humidity on the emissions of the diesel engines
      • 5552 / Vehicle drawbar test method with improved measurement and control
      • 6027 / Digital precision planning tool for autonomous forest regeneration of mixed tree species
      • 6630 / Real-time measurement of tire sinkage using stereo cameras
      • 6690 / Comparative evaluation of methods to evaluate penetration resistance for Clegg hammer and...
      • 6736 / Ride comfort comparison between 4-poster and straight line driving simulations
      • 6861 / Research on airplane performance on a grass airfield
      • 7198 / Analysis of aircraft dynamics while overcoming obstacles on a grass runway
      • 7477 / Evaluating the pressure performance of DPF filters using engine bench analysis
      • 7523 / Assessing performance of light wheeled vehicles on GRC-1 using 3D scanned footprint...
      • 7577 / Implementation of moving loads on ice in NRMM
      • 8151 / Study on estimation of traveling states using strain information on chassis of lunar and ...
      • 8254 / Initial steps towards characterization of a new military cold weather all-terrain vehicle...
      • 8672 / Trajectory optimization for vegetation override in off-road driving
      • 9268 / Experiments of sinkage characters of wire mesh wheel under different slip ratio
      • 9992 / Soil shear strength values obtained from its colour
    • Abstract-only submissions
  • Proceedings
    • Editorial Board
    • Book of Proceedings
  • Quick links
    • ISTVS
    • EasyChair
    • Journal of Terramechanics
Powered by GitBook

© International Society for Terrain-Vehicle Systems :: www.istvs.org

On this page
  1. Submissions
  2. Papers

1805 / Accelerating graph networks for real-time physics simulations

Paper presented at the 16th European-African Regional Conference of the ISTVS

Previous1638 / Analysis of tire characteristics on road surface with volcanic ash fallNext2145 / Obstacle performance and wheel failure test analysis of Zhurong rover

Last updated 1 year ago

Title: Accelerating graph networks for real-time physics simulations

Authors: Mohammad Jasim Usmani and Krzysztof Skonieczny

Abstract: Physics-based simulations are crucial for analyzing interactions between physical objects and are very commonly used in areas such as terramechanics. The discrete element method is an accurate way to model the interactions between the soil particles, however, it is very computationally intensive. One promising alternative is to use Graph Network architecture which uses graph neural networks to learn and approximate the dynamics of the physical system fed into it. Even though this is faster, due to the high dimensionality of the dataset, the inference is performed on a very large graph and rollouts are not in real-time. To accelerate the simulation, one way developed recently in our laboratory is to use dimensionality reduction techniques such as principal components analysis to identify the modes in the dataset that carry the highest variance. These modes are then fed into the Graph Network along with the rigid body to train the model. Although the reduced-order dataset results in a much smaller graph than the original dataset, the graph fed into the network is still naively fully connected. This research aims to further accelerate this subspace framework by identifying a partial graph for the network without compromising either the performance or training. This identification of the partial modes is done by Neural Relational Inference (NRI). The NRI model is based on a variational autoencoder where the encoder can extract a partial graph that can still perform inference on the GNS without loss in accuracy. We demonstrate the effectiveness of our approach on blade cutting as well as wheel simulation datasets and show that this framework has reduced the inference time from 0.35 seconds per second of simulation to 0.17 seconds, an approximately 2x speed up.

Order the full paper:

ISTVS members: receive three papers per year as part of your membership via the ISTVS Member Portal:

https://doi.org/10.56884/PFFS7901
https://www.istvs.org/proceedings-orders/paper
https://istvs.knack.com/member-portal/