In the world of vehicles, the automation of movement was always a task to be solved. For example, in a warehouse the automation of the forklifts could lower the expense of moving stocks, not to mention the safety of the package. The automation of certain tasks of the vehicle control is getting more common in the car industry as well. The main motivation for this is the safety of the passengers. The traction control system is responsible for preventing the driver from losing control of the vehicle. Or a lane keeping system prevents the vehicle from accidentally leaving its lane, and therefore lowering the risk of an accident.
In my thesis, I am investigating the behavior of fully autonomous car-like robots. For this, I will build a system, with which we can simulate an environment composed of obstacles, and the vehicle’s movement in this environment. Our goal is to mimic the behavior of a person-driven car. I will use path planner algorithms developed and implemented by the Department of Automation and Applied Informatics.
For defining the obstacles in the environment, I will build an environment editor. With the editor, we can define the problem to be solved by the path planner algorithms. The problem consists of the obstacles, which are polygons, and the vehicles start and goal states. These can be defined with the editor. The second task is the integration of the existing path planner algorithms into the system. To achieve this, I will use the services provided by the Robot Operating System. I will create a visualizer module, to display the environment and the vehicle’s path on a real surface.
The robot has to follow the path, that was created by the path planner. This will be achieved by a path follower algorithm. With this, we will be able to simulate the movement of the vehicle following its path in the editor software. For this algorithm, we need to measure the error of the vehicle’s position from the path. Then, we need to create an appropriate control signal for the robot, to be able to find, and then stay on the path. In reality, it’s likely, there is other moving vehicles in the environment. To be able to simulate this, we can define any number of robots in the environment editor. The path planner will create a path for each vehicle, and the robots will try to avoid colliding with each other on the move.
My goal is, to investigate the problems, and difficulties of the tasks of the creation of the environment, the integration of the path planner algorithms, the visualization of the virtual environment, and the path following, and to try to solve these.