Simulation Models
After some years of research in traffic flow and its applications on traffic planning and management, a wide variety of methods and tools has been developed. So, there are a large number of simulation models, usually characterised by the level of detail in which they describe the traffic processes.
In fact, each person has its own way of seeing things, its own point of view over a particular scenario. Thus, each analyst has his own perspective over a specific problem and needs to see through that point of view to be capable of analyse that problem. So, each simulation model is an abstraction of the domain that tries to represent a specific perspective for any specific kind of experts.
Below the different granularities that exist:
The different simulation granularities; from left to right: macroscopic, microscopic, nanoscopic (within the circle: mesoscopic)
Macroscopic models describe traffic at a high level of aggregation such as flows or densities. However, these kinds of models do not consider the constituent parts of flows such as the vehicles, which make them good for large and complex network analyses. They are also useful for route planning because they have a lower computation cost compared to the other models.
As an example, the Lighthill–Whitham–Richards (LWR) Model [1] use differential equations to formulate relationships among traffic flow density. These equations describe traffic flow density like flows in fluids or gases, and therefore, its solution can be obtained through simulation.
Microscopic models have a more detailed representation of the traffic than macroscopic ones. These models describe the behaviour of the entities that make up the traffic stream as well as their interactions. In microscopic models, the level of detail goes till the individual behaviour of vehicles, their interactions with each other and with the road network. For that, these models are capable of perceiving some rules of the vehicles’ behaviour such as when a vehicle accelerates, breaks, changes its lane and chooses/changes their routes to its destinations.
These types of models are widely used in analyses of detailed traffic situations such as traffic lights control. However, as it could be easily seen, this models have very high computation costs, and large/complex road networks are almost impossible to be simulated.
In the literature the Car-following model [2], Lane-change model [3], and Route-choice model [4] are the main methods used to determine vehicle’s behaviour.
Mesoscopic models fill the gap between macro and micro models. They normally describe traffic entities at a high level of detail, but their behaviour and interaction are in a lower level of detail. In mesoscopic models, vehicles can be grouped in packets, which are routed through the network and are treated as one entity.
A new trend of traffic simulation is the nanoscopic model which extends the capabilities of three basic components of microscopic simulation: vehicle modelling, vehicle movement modelling. and driver behaviour modelling.
It is mostly used in autonomous driving and a strictly relationship with automated robotic, because needs to simulate sensors and vehicles constitutes parts. Controls and great improves already has been done on this field. In the paper An Approach to Simulate Autonomous Vehicles in Urban Traffic Scenarios [5] is observed great potential in using robotic simulators on autonomous driving field, motivating an information exchange among robotic and traffic study groups.
[1] M.H. Lighthill and G.B. Whitham. On kinematic waves II: a theory of traffic flow on long crowded roads.
[2] J J Olstam and A Tapani. Comparison of Car-following models. Swedish National Road and Transport Research Institute, 2004.
[3] M E Ben-Akiva, C Choudhury, and T Toledo. Lane changing models. In Proceedings of the International Symposium of Transport Simulation, 2006.
[4] C G Prato. Route choice modeling: past, present and future research directions. Journal of Choice Modelling, 2(1):65–100, 2009.
[5] M C Figueiredo, R Rossetti, R Braga, and L P Reis. An approach to simulate autonomous vehicles in urban traffic scenarios. In Intelligent Transportation Systems, 2009. ITSC ’09. 12th International IEEE Conference on, pages 1–6, 2009.