How to make relevant a virtual reality situation?

Science topics February 2016 InnovationTransportModelling and computer simulationHuman behaviourRoad safety

Régis Lobjois, Researcher - COSYS department, LEPSIS Laboratory

When using driving simulators with the goal to investigate road user behaviours, we need to consider its validity. A simulator is valid when its users behave in a natural way without any preliminary training1. The main idea behind this is that the simulator should not lead to the adoption of new behaviour or strategies; otherwise it will fail to highlight the real user behaviour in the studied situation.

With regard to the validity of virtual reality devices, four key dimensions are generally taken into account2:

  • Physical validity, which relates to the properties and performance of the software and hardware architecture of the system, and hence its ability to provide sensory stimulations which are objectively close to those of the real world;
  • Subjective validity, which relates to a judgement of resemblance and credibility of the simulated situation;
  • Ethological validity, which gauges the similarity between behaviours in simulated and real situations;
  • Last, psychological validity, which relates to the similarity between the psychological processes underlying activity, including the mental load generated by the driving activity.

Evaluation of driving simulator validity

It consists of exposing users to the technical and technological choices, and hence to test the features they interact with and which immerse them in the virtual environment.

This testing is essentially based on subjective (by means of a post-exposure questionnaire or interview) and behavioural measures (selected speed, swerving from side to side, decision-making, etc.).

However, these two types of measures have limitations. Subjective measures are less sensitive to the characteristics of the virtual environment than so-called objective measures3. On the other hand, while the behavioural measures collected in a simulator can resemble those observed in real situations, the underlying processes may differ.

The approach we adopt when designing and evaluating simulators gives priority to what is known as the psychological dimension, by means of relative evaluations (comparing several devices with different characteristics) or absolute evaluations (comparing a device to the real situation). Here are some examples:

 

Motorcycle riding simulation

Because in simulation it is not possible to lean the “motorcycle” as a rider would do on the road (due to a feeling of imminent falling4), the visual scene is tilted in order to compensate for this limitation and create the illusion of leaning. However, when the visual scene roll angle is increased, researchers have shown that simulator sickness is stronger5. In addition, tilting the visual scene induces changes in the  visual strategies which may explain why the control of steering in bend is less smooth.

The psychological validity of tilting the visual scenery is therefore limited as it tends to modify the natural visual processes which are responsible for the locomotor control in bends6.

Moving base motorcycle simulator (top left) positioned in front of three screens with a height of 2.8m and a width of 1.8m (bottom left). The visual scenes can be rotated in the roll in order to create the illusion of tilting (images on the right, with and without visual tilting).
How to make relevant a virtual reality situation? - Ifsttar - Moving base motorcycle simulator

Road-crossing simulation

The improvements in tracking systems able to record pedestrian motion in a virtual environment has enabled LEPSIS researchers to compare the effect of two response conditions on road crossing decisions. In the first case, the user was simply asked to press a button to state if he would or would not cross the road in the presence of the traffic shown on the screen. In the second case, the participant was able to cross the road while walking over a distance that corresponded to the width of the simulated pavement. In the second situation, crossing behaviour were more opportunistic but less risky. This finding was explained by the fact that natural coupling between the perception of the situation and the control of the crossing action was better preserved in the second case than with the push-button type of response7.

 

 

Car driving simulation

Car driving simulators are mainly evaluated by comparing the driving performance and behaviour to the real situation. The question is therefore whether or not a given level of performance was achieved in both the real and simulated situations, with the same allocation of attentional resources. With this end in view, we have compared the mental workload generated by the two conditions and shown that mental workload was higher in driving simulation.

 

Taken together, these few findings show how important it is to take into account the psychological processes that underlie actions when we are trying to evaluate virtual reality devices.


Find out more...

1Burkhardt, J-M., B. Bardy, and D. Lourdeaux. 2003. Immersion, réalisme et présence dans la conception et l’évaluation des environnements virtuels. Psychologie Française, 48, 35‑42.

2Malaterre, G., and J. Fréchaux. 2002. Validité des simulateurs de conduite par comparaison de tâches réalisées en situation réelle et simulée. Actes INRETS n°82, pp. 149-157.

3Morice, A. H. P, I. A. Siegler, and B. G. Bardy. 2008. Action-perception patterns in virtual ball-bouncing: Combating system latency and tracking functional validity. Journal of Neuroscience Methods 169: 255‑266.

4Dagonneau, V. (2012). Etude des liens entre immersion et présence pour la mise au point d’un simulateur de conduite de deux-roues motorisé. Thèse de Doctorat, Université Paris-Sud.

5Lobjois, R., V. Dagonneau, and B. Isableu. (2016). The contribution of visual and proprioceptive information to the leaning sensation in a dynamic motorbike simulator. Ergonomics, DOI: 10.1080/00140139.2016.1149229

6Lobjois, R., I. A. Siegler, and F. Mars (2016). Effects of visual roll on steering control and gaze behavior in a motorcycle simulator. Transportation Research Part F, 38, 55–66.

7Lobjois, R., and V. Cavallo. 2009. The effects of aging on street-crossing behavior: from estimation to actual crossing. Accident Analysis & Prevention, 41, 259-267.