Enhancing tools for intelligent transportation systems applications: matching data acquired by driving simulator and travel simulator

Among ITS (Intelligent Transportation Systems) applications, a relevant interest has been devoted in recent years to ATIS (Advanced Travellers Information Systems). The prediction of travellers’ route choice, in a context in which they are provided with information by different systems (VMS; route guidance; in car navigation systems), includes the study of travellers’ compliance. In order to be effective, ATIS require appropriate levels of travellers’ compliance with the dispatched information (Bifulco et al., 2007). Compliance and accuracy of information systems are implicitly related. It is worth noting that an appropriate level of compliance (as well as of familiarity with the technologically dispatched information) is reached also (maybe mainly) with respect to recurrent conditions. Accuracy of information systems can be evaluated by respondent, on the base of the discrepancy between suggestions received by ATIS, and experienced travel times. Putting in place accurate ATIS is not only due to technological matters but also (and mainly) to modelling issues, mainly related to the fact that the ATIS-information design problem is, in recurrent traffic conditions and for dynamic and predictive ATIS, a typical anticipatory-route-guidance problem (Crittin et al., 2001). The design of an accurate information system is obtained not only by advanced technologies but also with complex implementations of iterative procedures. In order to reach highaccuracy performances, it is required the availability of proper simulation models where the effect of the information accuracy on the compliance with information is explicitly and endogenously modelled. In order to study and model the travellers’ response to the information systems, different approaches have been adopted for data acquisition. In particular, the difficulty in gathering data from the real world has induced many researchers to adopt the Stated Preferences approach. Several researches have carried out experiments by adopting computer-based tools (travel simulators) or by designing a virtual reality in a driving simulator (Klee et al.); in both cases the main advantage is identified by the possibility to control the experiment variables (e.g. actual costs; accuracy of information; the set of choices characteristics, etc.). Different kind of models in ATIS’ contexts have been already dealt with (Avineri, et al., 2003; Ben-Akiva et al. 1991; Emmerink et al., 1994; Ettema et al., 2006; Van der Mede et al., 1996), and the most common adopted simulators have been travel simulators; nevertheless, some studies have also been made by adopting driving simulators (Chang, H.L. et al., 2009; Katsikopoulos et al, 2000-2002). In a driving simulator (characterized by very expensive technologies) the experiment designed is more complicated with respect to the travel simulator; besides, the experiment takes more time with respect to the travel simulator. However, the main advantage of the experiments made by driving simulators is that the virtual and immersive reality induces a more realistic behaviour of the respondents. Driving simulators are particularly suitable when the focus of the experiments is on driving choices and are very efficient also in reproducing the mental workload induced by driving. Rarely, results obtained by driving simulator and travel simulator are compared (Bonsall et al.,2000; Katsikopoulos et al., 2000). This, in order to improve the design of information (in terms of accuracy and quality), on the base of respondents’ behaviour, and by identifying the biases introduced in the experiments (Koutsopoulos et al., 1995). In our work, two experiments have been made by adopting both a driving simulator (route choice virtual simulator) and a travel simulator (SP Platform- Bifulco et al., 2009). In the experiment the respondents’ reaction to variable message signs has been studied and modelled. Respondents have been provided with the same kind of information (mixed information – prescriptive plus descriptive) and at the same levels of accuracy (respondents have been tested in different scenarios, each of them characterised by a different level of accuracy). Therefore the running experiments is the same: respondents are asked to repeatedly make their choices. By coupling the web-based and the driving-simulator-based experimental contexts, two main experimental strategies for observing drivers behaviours have also been coupled