7.2 Realistic two-handed Haptic Manipulation
7.2.2 Interaction with complex Virtual Environment
In section 6.2.4, we performed the augmentation of a complex visual scene representing a small house. In this section, we present the results obtained (in terms of performance) when manipulating this environment.
The PC running the simulation is a based on an Intel Quad-Core processor (2.4 GHz) and a NVIDIA 8800GTX graphics card. When loading the simulation, we remark that between 75% and 100% of the microprocessor is occupied, which is as expected. At the beginning during few second, the program lags. The reason is that every dynamic objects are in the “moving state”, and that most of them are colliding. Then, once they stop, the refresh rate is stable. The Haptic Thread is running around 900 Hz. The Physics Engine has been limited to 300 Hz, and the display is also refreshed around 300 Hz. The visual result is presented on figure 7.6.
When immersed into a large scale Virtual Environment, the user is not able to reach every object. As he is seated in the Haptic WorkstationTM, it seems difficult to propose a
realistic displacement method. Thus, a metaphor should be used to let him move aground the Virtual Environment. In [41], authors presented an evaluation of 3 different techniques to achieve the interaction with objects that appear bigger than the workspace of the haptic device. We used a method similar to the “bubble” technique described in [40]. When the user moves his arms near the workspace limits, he enters into an area that displace the virtual camera: arms to the front moves the camera forward, arms to one side turns the camera. In addition, force feedback is applied to tell the user that he is entering to this area. This convenient method is intuitive, and the displacement is really controllable. However, we remark that a calibration has to be done to adapt the method to the user’s morphology. Indeed, the far limit of the workspace is not at the same position for everyone.
The force feedback is efficient with most of the static objects. We were surprised that even the thin tables produce a convincing resistive force due to the fact that the hand does not go through. The main limitation comes from objets augmented with pMaps. Even the static ones produce unexpected behaviors. It also occurs occasionally with dynamic objects that use convex meshes. Despite these issues, the feeling of being there (presence) seems increased, mainly because of the 3D interaction capabilities of the two-handed haptic device.
In this chapter, we presented applications with realistic interaction paradigms. In the first section, we proposed a study using an assembling training system under Mixed-Reality
7.2. REALISTIC TWO-HANDED HAPTIC MANIPULATION 101
conditions. We created a feasibility study that allows to perform a test on the efficiency with different kinds of realistic interfaces. We presented interesting results concerning precision, and proposed useful recommendations about the design of such systems. Then, in the second section, we showed some Virtual Reality applications made with MHaptic. The common point between these applications is that they are intuitive thanks to the two- handed haptic device.
Chapter 8
Unrealistic Human-Computer
Interactions
W
E SHOWEDin previous chapter that some applications, like Virtual Training, take advantage of realistic two-handed interactions. For an efficient training, we can distinguish the theory learning from the gain of experience. A two-handed interactive system can propose real life situations that increases the experience of the trainee. Some other applications do not simulate a real life interaction. They propose a paradigm based on metaphors, which is often the case in computer applications. The main disadvantage of this kind of interaction, is that it could require some kind of training and practice. But, when the experience is acquired, it could become really easy to perform complex tasks. In this chapter, we want to study these interaction paradigms, in order to see how a two-handed device could increase the learning speed without compromising the efficiency.As mentioned in subsection 2.2.1, teleoperation is historically the first field that took advantage of Haptics [61]. This is one of the reason that convinces us to evaluate this novel platform in this historical context. We propose to create a teleoperation platform that uses Virtual Reality techniques and a two-handed haptic device. We perform two experiments with two different categories of teleoperated robot: one for driving a small robot, and the other one with the remote control of a blimp. One may ask: “why two experiments with two robots ?”. In fact, we think that a specific interface for a specific robot is needed to maximize the efficiency in terms of remote control. And we are quite sure that a generic device, even with a high number of degrees of freedom, cannot perform better than a spe- cific interface. But, a generic device has the advantage to be easily reconfigurable to be adapted to many robots. This is the main characteristic that needs to be evaluated. The sections 8.1 and 8.2 present these two experiments.
8.1
Interaction Paradigm for Teleoperation
In this section we present the evaluation of different kinds of interfaces to remote control a small robot. We deal first with the technical description of the overall system, then we present our evaluation protocol, which is based on a robot race, and we finally discuss the results and conclude.