Autonomous driving is one of the most popular fields of research today in both the automotive industry and the
academic world. The main focus of this thesis is motion planning, a relatively small, but still huge
part of it (besides sensing and perception of the environment, situation analysis, decision,
The purpose of this thesis is to study motion planning algortihms, and propose feasible variants
for use in embedded systems, considering the limitations of the target platform. Since
general motion planning is incredibly complex, the scope is limited to highway driving.
After a short overview of the literature and theoretical background, A MATLAB/Simulink simulation
framework is introduced, which is capable of testing different planning algorithms, provides the
possibility to insert and move obstacles in their environment, and select the vehicle models used
by the planners. Two algorithms are compared, the Dynamic Window Approach (DWA), and the Rapidly
Exploring Random Trees (RRT), both of which have been tailored for the simulation environment, commercial
vehicles, and highway driving. It is shown, that the added benefits of the more complex planner
(RRT) are not always significant (if there are any), but can make some scenarios solvable, which
is unsolvable by the simpler planner (DWA).