The use of private vehicles is becoming more common each year, surpassing that of public transport [1], and currently, there are billions of vehicles on the roadways [2]. Despite ongoing technological advancements, road transport is still responsible for a number of problems that endanger both people and the environment. Nearly 65% of the consumption of worldwide oil-based products can be attributed to the transport sector [3]. Regarding emissions, road traffic is responsible for more than 20% of all greenhouse gas emissions in Europe [4], with carbon dioxide emissions rising yearly and significantly influencing climate change. Besides climate issues, road-transport-related emissions such as nitrogen oxides (NO_x) have serious negative health effects [5]. In this context, it is crucial to pay attention to how to encourage sustainability in the operation of these vehicles to quickly and effectively cut off the negative impacts of road transport, especially regarding emissions, which is aligned with the net-zero target of the European Commission and the United Nations 2030 Agenda of Sustainable Development Goals (SDG 3, 11, and 13). Since real-world measurement campaigns are usually not easy to deploy, either because of the increased risk of collision or the unnecessary production of pollutant emissions, the use of a driving simulator can be considered a good approach for conducting experiments by simulating a driving environment to study vehicle dynamics and driver behaviour on a microscopic level. Driving simulators accurately simulate the driving experience without putting the driver or the environment at risk [6].

This exploratory analysis microscopically assesses how well a driving simulator might reflect the effects of vehicle emissions through data related to vehicle dynamics.  This study aims to validate and compare the results of tests performed in both real-world and simulation environments. The real-world and simulation tests are designed to demonstrate not only the variations in vehicle emissions but also the variations in drivers’ behaviours, considering different types of routes. The driving simulator software and the On-Board Diagnosys (OBD) system coupled with the Global Navigation Satellite System (GNSS) provide instantaneous speed, acceleration, and position data for the vehicle, among other variables, under the simulation and real-world environments, which are carefully analysed.

Driving simulator experiments present different advantages for research purposes. Driving simulators have good controllability, reproducibility, and standardisation and allow the possibility of simulating dangerous driving conditions without putting the driver at risk ^[7][8][7,8]. Regarding the data collection process, a driving simulator can measure performance accurately and efficiently, while it is far more complex to obtain complete, synchronized, and accurate measurement data from a real vehicle [7].

A considerable number of the scientific studies using these tools are framed in the clinical and psychological context ^[9][10][9,10]. A study conducted in 2021 used a driving simulator to determine reasonable speed limits for safety in dynamic low-visibility foggy conditions [11]. Participants completed trials with varying visibility and speed levels. A quantitative model was established and suggested speed limits were proposed based on visibility changes. The findings inform the development of variable speed limit (VSL) systems and reduce crash risks in foggy conditions with poor visibility. More recently, a study was conducted that emphasises the criticality of analysing driving behaviour to evaluate road safety, emissions, and fuel consumption [12]. The research takes into account variables like traffic conditions, road characteristics, and driving profiles. Through driving simulation tests, the study investigates how driver–road interaction influences gear-shift strategy, vehicle dynamics, safety indicators, comfort variables, and pollutant emissions. Performance assessment and database creation support in-depth analysis. Another research study that falls into this category is the 2018 study that aimed to address human errors and improve driver behaviour on curves by using different road-marking treatments [13]. Two treatments, optical circles and herringbone patterns, were tested in a driving simulator experiment on rural road sections. The study concluded that optical circles are effective for speed reduction and increasing driver attention, while herringbone patterns can help prevent head-on crashes by improving lateral position.

While the use of simulators has gained traction as a valuable tool in various domains, their effectiveness in reproducing real-world observable driving behaviour is not always demonstrated or evaluated. Therefore, validation is one of the most important topics to be studied in the driving simulation since the viability of all driving simulation studies depends on it.

Validity is the capacity of a simulator to simulate actual driving reliably [14]. The validity of driving simulators may be categorized into absolute validity and relative validity. Absolute validity is recognized when there are no statistical differences between driving behaviour measures observed in the simulated world and the real world ^[14][15][14,15].

The relative validity of the driving simulator should be assessed in situations when absolute validity is not achievable. Relative validity describes the extent to which the variation of a factor in a simulated world has an identical influence in the real world [16]. For instance, a research study can comprise 2 (driving simulator and real-world) × 2 (sober and drunk) designs for studying the behavioural change in the driving parameters. Here, researchers might find statistical differences in the numerical values observed in a real and a simulated world. However, relative validity can be achieved if the main effect of alcohol on different driving behaviour parameters shows similar effects in real and simulated worlds [14].

Meuleners and Fraser [17] focused on validating a laboratory-based driving simulator and provided early support for the relative validity of the driving simulator. The researchers recruited 47 drivers with valid licenses and instructed them to drive a specific route both on-road and in the driving simulator. The driving behaviours of the participants were assessed by an occupational therapist and two trained researchers using an assessment form. The results showed no statistical difference between the on-road assessment and the driving simulator for various driving behaviours, including mirror checking, observations, speed at intersections, maintaining speed, and obeying traffic lights and stop signs. These findings indicate that the driving simulator has relative validity and can be utilized for various road safety outcomes, providing a safer alternative to on-road testing and reducing the risk of harm to participants.

In [18], a general procedure was conducted for validating a driving simulation environment to analyse gap acceptance behaviour. The authors tested whether a synthetic indicator of gap acceptance behaviour showed significant differences when computed based on the simulated environment versus empirical observations. The proposed validation procedure was applied to the case of a three-leg roundabout. The results show that the mean critical gap estimated in the field and the mean critical gap estimated in the simulation environment were not significantly different. The proposed procedure can be applied in various contexts where gap acceptance behaviour is a central element in terms of the safety and operational performance of the traffic system under analysis.

Other studies were devoted to exploring the relevance of the vehicle specific power (VSP) mode distributions resulting from the empirical and simulated trips ^[19][20][19,20]. Zhao et al. developed an eco-driving feedback system using a driving simulator to enhance eco-driving training [20]. The system provides real-time feedback and voice prompts during driving to improve drivers’ eco-friendly behaviour. After driving, participants receive an evaluation report with fuel consumption rank, potential fuel savings, and personalized driving advice. The researchers used a microscopic emissions model based on VSP distribution to calculate the emissions on the empirical trips. In testing, the system led to a 5.37% reduction in CO₂ emissions and a 5.45% decrease in fuel consumption. These findings highlight the system’s effectiveness in promoting eco-driving behaviour, resulting in lower emissions and fuel usage. In 2016, Yu et al. proposed to test and validate the feasibility and applicability of a driving simulator approach in generating vehicle activity data to produce VSP values and then to estimate emissions [19]. The study concluded that the driving simulator can be considered a useful test tool for estimating vehicle emissions, particularly for scenarios where the driving time is relatively short and the network and traffic conditions are less complex.