Autonomous Driving Simulation Framework for Traffic Studies

Simulation of autonomous connected vehicles for driving policy design in Mcity.

An intelligent transportation systems (ITS) is a typical cyber-physical system (CPS) in which physical components, for example, Connected Automated Vehicles (CAVs), are monitored and controlled through a network of cyber and physical components. Such systems, therefore, contain event-driven dynamics along with time-driven dynamics. The proposed discrete-event and hybrid simulation framework based on SimEvents facilitates testing for safety and performance evaluation of an ITS and has been used to build a traffic simulation model of the Mcity test facility. It is specifically designed for testing CAVs and contains various road/lane configurations and a complete instrumentation system. This enables users to study traffic at the microscopic level, including the effectiveness of new control algorithms for CAVs under different traffic scenarios, the event-driven aspects of transportation systems, and the effects of communication delays. The framework spans multiple toolboxes including MATLAB, Simulink, and SimEvents.

Three major different scenarios are included:
1. CAVs crossing two adjacent unsignalized intersections using optimal control
2. CAVs merging at freeway on-ramp using (i) optimal control (ii) control barrier functions
3. CAVs and/or non-CAVs crossing a signalized intersection using optimal control (CAVs) and the intelligent driver model (non-CAVs)
By specifying the Block Parameters of the MATLAB Discrete-Event System (MATLAB DES), you may try different controllers given a particular scenario. For example, for CAVs merging at freeway on-ramp, you can select either an optimal controller or a control barrier function-based controller. In addition, you may observe the queue length of the signalized intersection 2 if you'd like to study the queueing behavior. Feel free to make edits or create your own traffic scenarios depending on your research purpose.

NOTE:
1. When running simulation for a certain scenario, e.g., signalized intersection 1, please set the MATLAB Discrete-Event System blocks of other scenarios, e.g., unsignalized intersections, freeway on-ramp, etc., to "xxx_Dummy" mode.
2. Double click the initialization block "init" if the CAV and INFO entities are not load automatically

Visualization is primitive. Ongoing work includes improving the visualization and creating a map for the whole Mcity.

Demos can be found on github (https://github.com/nebneBgnahZ/mathworks_autonomous_driving_project) or Youtube:

1. CAVs crossing an unsignalized intersection using optimal control: https://youtu.be/jhVzNrTySnA

2. CAVs crossing an unsignalized intersection with turns using optimal control: https://youtu.be/RXdoaW9asAg

3. CAVs crossing an unsignalized intersection with turns using control barrier functions: https://youtu.be/UH0tJ5imwoU

4. CAVs crossing a signalized intersection using optimal control: https://youtu.be/WkEODHh6smM

5. CAVs merging at freeway on-ramp using optimal control: https://youtu.be/bcYiErbJCSM

6. CAVs merging at freeway on-ramp using control barrier functions: https://youtu.be/szMFeqnpXrc

7. CAVs crossing two adjacent unsignalized intersections using optimal control: https://youtu.be/1LK-7mR_OBg

8. Non-CAVs crossing a signalized intersection using intelligent driver model: https://youtu.be/dUBoj9O7RvE

9. [Mixed traffic] CAVs and non-CAVs crossing a signalized intersection: https://youtu.be/on6e9fyn084