Lidar Guided Navigation on a Differential Wheeled Robot

This thesis is about guiding a simple differential wheeled robot using lidar. Due to its simple construction, the movements of the robot are rather inaccurate. The objective is to guide the robot towards a target position using the data provided by the lidar. Therefore, the properties of the lidar are outlined and a series of lidar test measurements are conducted. Valuable insights are gained from this series of measurements, such as the extent of faulty measurements and the effect of interpolation methods to bridge the faulty measurements. During this thesis, two types of algorithms are proposed to estimate the rotation of the robot. The first algorithm utilizes functions similar to the correlation coefficient, whereas the second algorithm utilizes functions similar to the cross-correlation. Both algorithms are examined regarding their complexity and accuracy.
Due to its lower complexity and slightly better performance, the second algorithm is implemented on an FPGA. The robot control logic and the communication with Matlab is handled by an Arduino Nano 33 BLE. Furthermore, a simple distance estimator based on the median is implemented on the Arduino as well. Moreover, a GUI is implemented in Matlab, which illustrates the environment surrounding the robot and the movement target positions can be specified in the GUI as well. Finally, the accuracy of the implementations is evaluated.