Slope Detection and Straight Line Segments based EKF SLAM

It is envisioned that in the near future mobile robots will be assisting human beings in their daily lives. Such robots are designed and developed to be intelligent enough to operate and share the same workplace with ordinary people. In such situation, robots are expected to be intelligent enough to move autonomously, quickly and safely without human’s control, in another word, fully autonomous navigation. In order to realize the autonomous navigation, the robot needs to acquire the model of the surroundings and estimate the own pose with respect to the environment. This problem is well known as Simultaneous Localization and Mapping problem, which is commonly abbreviated as SLAM problem, has been investigated by many researchers since it is a basic requirement for realizing effective autonomous robotic navigation and operation. SLAM problems arise when the robot does not have access to a map of the environment, nor does it have access to its own poses. Instead, all it is given are the sensors measurements and control inputs. The goal of the SLAM problem is to reconstruct the map of the environment and estimate the pose of the robot simultaneously. SLAM problem is the problem of spatial exploration, and it could be described as a Chicken and Egg problem. It would be a simple problem to localize the robot if the true map of the environment were available. Similarly, if the pose of the robot is given, constructing the map of the environment would be an easy job. However, it becomes extremely difficult when the localization and mapping are implemented simultaneously. Feature based maps are widely adopted in various SLAM research works for its compactness. The drawback of the feature based maps is that it suits to the structured environment only. In this thesis, we consider line segment based representation because it requires very little memory and numerous line segments exist in a typical indoor environment when we use the laser range finder (LRF) to scan the environment.