Microscopic Traffic Simulations for Exploring the Interaction Dynamics of Connected & Autonomous Vehicles in Merging Scenarios
Examensarbete för masterexamen
As the number of Autonomous Vehicles (AVs) in today’s traffic increases, it is important to incorporate their existence in traffic simulations to consider their possible negative effects on traffic and predict their advantages and disadvantages to propose feasible solutions concerning their deployment and operation. In this regard, having a realistic traffic simulation increases the accuracy of such prediction results. Therefore, the main aim of this master thesis is to design realistic traffic simulation environment for mixed traffic including AVs, to analyze the interaction of AVs and Human Operated Vehicles (HOVs). At the same time, analyze the AVs effects on traffic flow in terms of efficiency and safety. To achieve this goal, the thesis studies the background theory and terms behind traffic simulation and safety indicators, it also studies the traditional car-following and lane-change models which are the fundamentals of the traffic simulation. Then, in order to simulate the interaction patterns of AVs with HOVs and eventually examine their effect on traffic safety and flow parameters, this thesis uses Simulation of Urban Mobility (SUMO) as the microscopic traffic simulation tool. Also, in second half of thesis, SUMO is integrated to Unity 3D game engine to create 3D model of the original 2D scenario in SUMO and control the ego vehicle with human driver as live agent which can make our simulation more realistic. Besides, it allows the interactive agent behavior to be represented in contrast to ’reactive’ agent structure of SUMO. This also gives us the opportunity to go beyond the ’omniscient’ and centralized structure of SUMO. For this reason, TCP/IP connection have been used to generate a communication between SUMO and Unity, and SUMO-TraCI library have been used to transmit vehicles and other simulation information between these two platforms. As a case study for human-in-the-loop co-simulations, we focused on on-ramp merging scenario. The aim of this case-study is to understand what kind of effects AVs will have on the number of safety critical events (SCE) and traffic flow in a merging scenario. For this reason we consider three possible main scenarios, (1) normal connection of on-ramp road to main road, (2) merging in presence of acceleration lane in merge zone, (3) connected AVs in merging areas. Additionally, each scenario is divided to several sub-scenarios which considers different combinations of numbers for AV and HOV during merging at on-ramp. Finally, few ADAS (Advanced Driving Assistance System) features have been added to simulation environment to analyze the impact of multi-cue interfaces on drivers. For this purpose, two types of cues have been designed in Unity 3D environment, visual and auditory interfaces. During designing these interfaces the Time to Collision (TTC) and distance to lead vehicle have been considered as safety indicators. (1) visual interface: the driver being warned by visual interface with appearance of different colors during driving, like red represents dangerous, yellow represents the situations needing attention from driver. (2) auditory feedback: by setting different sounds to represent the different situations of driving into the simulation environment, the driver has been warned.
Traffic simulations , heterogeneous vehicle fleet , autonomous and connected vehicles , merging scenario , safety analysis