IEEE Access (Jan 2024)

Low-Cost Modeling and Haptic Rendering for Membrane-Like Deformable Objects in Model-Mediated Teleoperation

  • Yishuang Sun,
  • Fanle Meng,
  • Dong Yang,
  • Mengchen Xiong,
  • Xiao Xu

DOI
https://doi.org/10.1109/ACCESS.2024.3467706
Journal volume & issue
Vol. 12
pp. 141198 – 141210

Abstract

Read online

We study low-cost modeling and haptic rendering methods for deformable objects in model-mediated teleoperation (MMT). An accurate local model is key to achieve good perceived realism in MMT. Depending on the materials of the objects with which the haptic device’s end-effector interacts, the deformation process exhibits specific features. In previous work, a point cloud-based virtual environment was built as the local model. While this model can represent the deformed surface, it is overly simplistic (mainly for rigid objects) and cannot capture the specific features of elastic membrane materials. On the other hand, classic finite element model can provide relatively more accurate dynamics, but cannot reach the 1000 Hz haptic rendering requirement due to its high computational complexity, leading to force instabilities. These limitations make the local model inaccurate and unable to provide precise haptic and visual feedback to the human operator. This work focuses on optimizing the MMT system for handling elastic materials. It primarily concentrates on extending the existing deformation algorithm for elastic membranes with oscillation and significant inertia. The first step is to analyze and simulate the membrane dynamics under force using wave equations. The second step is to implement the modified deformation algorithm to create a new point cloud model with significantly reduced computational complexity, thus achieving the 1000 Hz haptic rendering requirement. This is to use a wave function to replace complex dynamic models. It allows for the simultaneous calculation of position changes for all points on the surface, rather than recalculating the deformation of each element individually as required in FEM models. Finally, an independent experiment and a real scene-simulating experiment are conducted to evaluate the performance of the proposed point cloud model and to simulate the real scene in MMT. Experimental results show that human operators can subjectively experience a realistic haptic and visual feedback when interacting with elastic materials via the MMT system.

Keywords