TRAFFIC MODELLING OF SELF-DRIVING VEHICLES
Søren Frost, COWI
Analysis of self-driving vehicles on the motorway system by simulations in VISSIM. What is the effect on road capacity and road geometry. Can simulations i VISSIM be used as a tool for analysing self-driving vehicles
In the last 2 year, COWI has done a lot of different analysis of traffic modelling of self-driving cars.
The Danish Road Directorate wanted to assess the extent to which self-driving vehicles will be able to increase the capacity of the motorway network and thus reduce the need for road extensions. In this context, the Danish Road Directorate considered if simulation in VISSIM of self-driving vehicles can be used as a tool.
The perspective by carrying out simulations of self-driving vehicles is to examine to what extent the simulation of self-driving vehicles may be used in the decision making process of expansion of the motorway network.
The simulations should be seen as the first attempts to use simulations in VISSIM as a tool to analyse self-driving cars. The analysis has been limited to examining the overall view and whether simulation can be used for analyses of self-driving cars on the motorway network.
At the beginning of the analysis, the main task was to limit the analysis so that it could provide a clear picture of whether simulations in VISSIM can be used to simulate self-driving vehicles. There are many unresolved issues in the self-driving vehicles, such as the vehicle character, size, power, drive ability as acceleration/deceleration, gasoline/ diesel/electricity/gas and the vehicles behaviour on the roads. Many questions, but at present are few answers to the questions.
The first analysis of the self-driving vehicles was carried out for a section of Holbaekmotorvejen near Copenhagen. A simulation model of the section was built in VISSIM. The section included several ramps and sections with both 3 and 4 lanes. Only a few assumptions about the self-driving vehicles compared to traditional vehicles were incorporated, namely self-driving vehicles drives closer to the vehicle in front and the self-driving vehicles have a smaller variation in speed than traditional vehicles. The share of self-driving vehicles were analysed with varying proportions of the total traffic volume. To make it possible to compare with the situation without self-driving vehicles, simulations of the current geometry with traffic levels for 2014 and future traffic in 2030 where both a low and high traffic was carried out. In alternatives where self-driving vehicles was introduced, dedicated lanes for the self-driving cars was established. In the simulations were calculated results for travel times, delay, speed, emissions, fuel consumption and capabilities.
The second analysis of simulations the self-driving vehicles in VISSIM was carried out for two sections on the motorway on Funen. One section with 2 lanes and a section with 3 lanes. As in the first analysis the driving behaviour and speed, distribution for the self-driving vehicles was changed compared to traditional vehicles. The main focus of the second analysis was on capacity and road geometry. How will the introduction of self-driving vehicles on motorways increase the capacity and shall the road geometry be changed to narrower lanes and/or dedicated lanes. The simulations was done for different traffic levels, share of self-driving vehicles, number of dedicated lanes and changed width of the lanes.
The analysis showed that the simulations in VISSIM could be used as a tool to assess the extent to which self-driving vehicles can increase the capacity of the motorway network and thus reduce the need for extensions. The simulations also showed the influence on the road capacity and road geometry for different motorways, traffic levels and share of self-driving vehicles.
However, the analysis also showed that there are still many unanswered questions about self-driving vehicles. The analysis responded to several questions about the self-driving vehicles, but there were also many new questions in the analysis.
Further analysis with simulation of self-driving vehicles could be:
Various parameter for self-driving vehicles behaviour, such as lane change behaviour, car following model, vehicle size, acceleration/deceleration, fuel type and speed profile?
Placement of self-driving trucks in the road network?
Do all self-driving vehicles has same property or there are differences between generations of the self-driving vehicles?
The interaction between self-driving vehicles and traditional vehicles cars in merging areas. And how big effect will this have on the capacity?
Can track width be reduced as a result of self-driving vehicles, which can lead to an extra track on the existing road space and thereby higher capacity?
Association for European Transport