Rendering 6-DOF Object-to-Object Interaction with 3-DOF Haptic Interfaces

Lead: Lionel Lee
2013-2014 (MSc Project, Completed)

Abstract of MSC thesis

Three degree-of-freedom (3-DOF) tool-based haptic interfaces are widely used in virtual environments as an affordable way to train operators, as well as for virtual prototyping and design. In some special cases with higher requirements on haptic fidelity, the tool needs to be modeled as an object with real volume rather than a single point. However, such object-to-object interaction will inherently involve reaction torques, which 3-DOF haptic interfaces are incapable of rendering. As a result,whenever reaction torques are induced, undesired system behavior will occur, such as rendering errors, vibrations or even instability. Six degree-of-freedom (6-DOF) interfaces with torque feedback would be a solution of this problem, but these devices are costly, hard to maintain, and would require more computational power to determine the feedback. This paper presents a penalty-based algorithm to realize stable yet convincing object-to-object interaction with 3-DOF haptic interfaces. The major contribution of this work is the regulation of excessive directional combined stiffness when multiple contact points are considered in the calculation of rendered force. In contrast to other 3-DOF rendering methods, this approach is to generate translational movement to resemble the dynamics of end-effector during torque-involved interaction. A virtual peg-in-hole task was conducted to evaluate the performance of the proposed algorithm. We used the geometrical constraints to calculate an ideal trajectory of the end-effector as a function of the peg’s orientation. The result shows that the end-effector’s trajectory resembled the ideal one when the virtual tool was rotated in the hole. We also showed that the regulated combined stiffness converged to a desired value so that the system stayed stable throughout the whole interaction.

MSc project


Associated Research Programme:
Involved Companies:

Piet Lammertse (MOOG) was co-supervisor for this project.

Created by The Laboratory Network