During my internship at dSPACE I developed a Lane Centering and lane change decision making algorithm using dSPACE’s SIL tools. The controllers were developed in Simulink and tested with dSPACE MotionDesk, dSPACE VEOS, dSPACE ModelDesk, and dSPACE ControlDesk.

The lane centering controller was based on an the Stanley Controller used by Stanford during the DARPA Grand Challenge. I added a speed-dependent dynamic look-ahead distance and tuned the controller on a set of different virtual test tracks.

The lane change decision making policy is a rule-based algorithm. There is no separate lane change controller. The vehicle changes the lane simply by switching to the reference polynomial of the adjacent lane.

The vehicle uses the data from virtual sensors to react to the lane lines, cones, traffic signs, and other vehicles.The demo video shows the algorithms performing on one of the test scenarios.