On 03/23/10, the students of the Software Engineering Lab
presented their solution in a final presentation. The goal was to model and implement an autonomous car on top of a given traffic simulator. The car should be able to drive from an arbitrary position A to another arbitrary position B without producing any accidents with other cars. Major sub-problems of this task were the implementation of a dynamic route finding algorithm and routines for safely overtaking other cars. Furthermore, network messages could be exchanged between different cars via a simulated wireless communication interface, for instance to detect other cars from greater distance. Hereby, the students were able to get some insight in current research topics in the field of car-to-car (Car2Car)-communications as well as to gain experience in the field of discrete event-based simulation. From a software engineering perspective, the development was based on the waterfall process. Also, the students were given the chance to put theoretical knowledge gained in other software engineering lectures into practice.
The solution implemented by the students is called "Clever Autonomous Driving Implementation" (CADI). It comes with a real-time visualization component that is accessible through a network interface. This way it is possible to observe the very same simulation on more than just one computer at the same time.
The autonomous driver's behavior is shown below with the help of two exemplary scenarios. The videos also illustrate the visualization component developed by the students.
For more information please contact Konrad Jünemann
, Natalya An
or Jochen Dinger