System wykrywania przeszkód łazika Scout bazujący na stereoskopowej kamerze głębi Obstacle detection system for the Scout Rover based on depth stereo vision

Kurzfassung

The master's thesis concerns project of software that implements various algorithms to obtain improved depth maps and perform their analysis. Filtration methods used allow for manipulation of depth data and segmentation of objects to detect obstacles. Estimation of their features has been implemented to aid further Scout Rover vision system development. Main tool used uses ROS framework as well as PCL library to manipulate data obtained with RealSense D435i camera. The thesis is built around five sections. Initial one provides brief introduction as well as background for motivation of research. Second chapter on describes topics concerning filtration, segmentation, and feature extraction tasks, as well as literature examples and possible solutions. Following chapter presents implemented methods which leads to next chapter that describes results. Last chapter includes conclusions, issues, and future research ideas.