Robotics is the most famous branch of electrical engineering. More advanced robots are being made with the improvement of the computing capacity, and they are being used in more occasions. In the 21st century the autonomus and self-driving robots are gaining popularity, which need information about the environment and their locations in it.
The mapping and localization can't be separated, because they are depending on each other, hence it is referred as the "chicken or egg" problem. In the previous 30 years lots of studies and researches have been done about this topic.
In my thesis I am going to go through the modern solutions of the SLAM problem, and explain them. With this knowledge, I am going to implement mapping and localization on a car model with GMapping, which is based on the Rao-Blackwellized particle filter.
Since the 2-dimension sensors often don't provide enough information, I am going to map the environment in 3-dimension with the help of a laserscanner and a tilting platform. The obtained information will be represented as a pointcloud, and I will use it that way in the future tasks.
During my tasks, I found an optimal parameter set for the Gmapping algorithm. I fitted the 3-dimension pointcloud on that map, and integrated these new information into the collision-avoidance algorithm.