Multiple Autonomous Vehicles In Complex Scenarios
Project course in automatic control.
Emulate complex traffic scenarios using both physical and simulated cars. One of the physical cars act as the ego vehicle whilst all other vehicles drive with a constant speed in a predefined lane. The ego vehicle consists of a planner & control module, perception module and behavior layer.
The planner is quite rudamentary in that each vehicle has full knownledge of the map. The map is divided into segments consisting of discrete points that are then fed to a pure pursuit controller.
The perception module consists of an extended kalman filter that utilizes the vehicles imu for state estimation when position measurements from Qualisys is missing. The module also consists of functionality for map creation and Lidar.
Given knownledge about position and speed of cars in the ego vehicles proximity, make decisions on whether overtaking is feasible. If there is a risk of collision, the vehicle will regulate its speed in order to follow the car ahead until a gap is found.
The visualization is done both in the ROS package Rviz and in a Matlab figure window. The Matlab figure is projected on the floor and is used to visualize the virtual cars and the planned and traveled path of the ego vehicle.