Diseño de la estrategia multidimensional

The basic idea is to make the automated vehicle’s near future path visible to the driver. This is similar to a so-called flight path vector (FPV) as used in aircraft head-up displays. As a minimal requirement, we assume a semi-automated system that is able to maintain the speed and keep the vehicle inside its definite lane. Extracted from sensor and camera data, an automated control system has an internal model of the road ahead (e.g. the next three seconds in the future) and computes a projected trajectory which the controller follows. We make this trajectory visible (cf. Fig. 5.1). Thus, in theory, the driver can recognise a deviation from the intended course potentially earlier than without support. The natural reaction – turning the steering wheel in

the opposite direction – is located on Rasmussen’s skill- or rule-based level (cf. Fig. 2.1). The concept of the prospective driving path display is based on Augmented Reality (AR) which means that virtual information is added to the real world. In our case, we add the projected future vehicle path virtually to the real road view. The path lies like a carpet on the road in front of the vehicle and moves itself with the car, which is why we call it theMagic Carpet. The concept has been filed for patent [152]. For simplicity, we will call the driving path displaycarpetfrom now on.

road surface

end of driving path

driving path

Figure 5.1: Sketch of a prospective driving path display. The internal representation of the vehicle’s projected guidance trajectory is mapped onto the road in form of a virtual coloured area. The vehicle will take the projected path and in a few seconds reach the position where the end of the driving path currently is.

AR is a paradigm within the field of computer science and can be described as a computer- supported augmentation of real world elements with virtual elements. The term AR was coined in the early 1990’s and was first classified by Milgram’s Virtuality Continuum, with the real environment on the one end and the virtual environment on the other [109], cf. Fig. 5.2.

Although AR is not limited to a single sensory channel, most applications only make use of visual content. AR does not replace, but rather enriches reality by adding information and creating a composite image of the real world and virtual objects. This can be achieved using different techniques. A common way of realising AR applications is the use of head- mounted displays (HMDs) showing a merged image of real and virtual content. The virtual elements are naturally only visible to the person wearing the HMD. Another approach is using a handheld display as a "magic lense". Virtual objects are overlaid onto a camera image and shown on the display. Today, mobile phones equipped with a camera are the standard platform for handheld AR applications, since they have become powerful enough to perform the

Real

Environment

Virtual

Environment

Augmented

Reality (AR)

Augmented

Virtuality (AV)

Figure 5.2: Milgram’s Reality-Virtuality continuum (redrawn from [109]).

necessary image processing routines. It is also possible to project virtual elements on two- or three-dimensional surfaces in the environment or to use 3D displays as an output medium for AR.

There are many examples of useful AR applications from various domains, but we want to focus on AR for automotive uses. For the use of AR inside a car, there is another technique available which can display images directly overlaid on reality through mirroring the virtual content in the windowpanes in a perspectively correct way (cf. Fig. 5.3). This technique is called

contact analogue and is commonly realised by advanced head-up displays. Thus, the virtual

image appears to be in fact merged with reality. A popular application using this technology is contact analogue navigation (e.g. [56, 92, 116, 180]) that does not only give navigational hints, but shows the driver which way must be taken on the road. A slightly different approach is the

virtual cable1, that appears as a red line high above the road to lead the driver the way.

Figure 5.3: Augmented reality navigation. Left: Extra display (from [116]). Right: Contact analogue display (from [143]).

Apart from navigation applications, there are several examples of automotive AR that support driving safety. As early as in the 1970’s Bubb presented a contact analogue concept that displayed the safety margin in the form of a brake bar. The brake bar indicated the point

on the road where the vehicle would come to a stop with full braking force applied [23]. This concept has been taken up by Tönnis et al. (cf. Fig. 5.4, [160, 161]), also including the distance to other vehicles. Bergmeier explored the potential of contact analogue marking of pedestrians in the dark using a night-vision system [8]. As Fig. 5.3 and Fig. 5.4 show, such systems are

Figure 5.4: Contact analogue brake bar indicating the stopping distance (from [161]).

mostly in a prototype stage, not yet of high quality and not yet commercially available. Contact analogue displays are technically complex, require large installation space and suffer from some intrinsic problems. Inappropriate overlays are still an unsolved issue. When driving closely behind another vehicle, a virtual navigation path, for example, would be shown overlaid on the vehicle, not on the road. Moreover, not all desired content can be shown, due to limited display area (cf. [180]). Israel and Bubb show an overview of the potential and limitations of the contact analogue HUD in vehicles [83]. Stanton and Pinto present a study showing effects of risk homeostasis when using vision enhancement systems [155]. That means, drivers tend to compensate the gain of safety and comfort achieved through a supporting system by driving more risky.