Simultaneous Localization and Mapping with Limited Sensing Using Extended Kalman Filter and Hough Transform

Original scientific paper The problem of a robot to create a map of an unknown environment while correcting its own position based on the same map and sensor data is called Simultaneous Localization and Mapping problem. As the accuracy and precision of the sensors have an important role in this problem, most of the proposed systems include the usage of high cost laser range sensors, and relatively newer and cheaper RGB-D cameras. Laser range sensors are too expensive for some implementations, and RGB-D cameras bring high power, CPU or communication requirements to process data on-board or on a PC. In order to build a low-cost robot it is more appropriate to use low-cost sensors (like infrared and sonar). In this study it is aimed to create a map of an unknown environment using a low cost robot, Extended Kalman Filter and linear features like walls and furniture. A loop closing approach is also proposed here. Experiments are performed in Webots simulation environment.