Network-based Haptic Systems with Time-Delays
During the last decades, virtual environments have become very popular and are largely used in many domains as, for example, prototyping, trainings for different devices, assistance in completing difficult tasks, etc. The interaction with the virtual reality, as well as the feedback force, is assure...
| Main Author: | |
|---|---|
| Language: | ENG |
| Published: |
Supélec
2012
|
| Subjects: | |
| Online Access: | http://tel.archives-ouvertes.fr/tel-00771948 http://tel.archives-ouvertes.fr/docs/00/77/19/48/PDF/Liacu_Bogdan_These_VF.pdf |
| Summary: | During the last decades, virtual environments have become very popular and are largely used in many domains as, for example, prototyping, trainings for different devices, assistance in completing difficult tasks, etc. The interaction with the virtual reality, as well as the feedback force, is assured by haptic interfaces. Generally, such systems are affected by communication and processing time-delays, resulting in a deterioration of performances. In this thesis, a complete study of the existing methods, as well as theoretical tools and new solutions, are proposed for the haptic framework. First, a comparative study, based on the experimental results obtained on a 1-dof haptic system, highlights the advantages and drawbacks of the most common control algorithms ported from teleoperation to haptics. Next, the theoretical tools needed in analyzing the stability of the delayed systems in different situations, as well as the physical limitations of the experimental platforms considered, are examined. Besides the standard case of constant time-delays, uncertainties are also considered and modeled by different types of distributions (uniform, normal and gamma distribution with gap). In the sequel, for overcoming the drawback of time-delays, two new approaches are proposed. First, the use of Smith predictor-based control is addressed and a specific solution for haptic systems is developed and discussed. The main idea is to introduce into the Smith predictor the environmental forces by using the additional information from the virtual reality regarding the distances between the controlled virtual object and other objects in the scene. To overcome the loss of performances induced by using a fixed gain in the controllers for all situations (free or restricted motions), the second approach proposes a gain-scheduling Proportional Derivative control strategy depending on the distance until a possible collision. Both approaches are experimentally validated on a 3-dof haptic platform, under different scenarios elaborated gradually from simple situations - free and restricted motion, contacts with moving objects, to more complex situations - virtual box with fixed or moving sides. |
|---|