This MSc thesis presents a solution for the camera-based visual navigation of a mobile robot. According to the specifications, a navigation method based on passive markers allowing easy reconfiguration has been established. The mobile robot used is the NI Robotics Starter Kit II platform equipped with a low-cost camera module. Software components of the image processing and navigation tasks were implemented in LabVIEW environment.
Exploiting the features of the processor card the image processing modules are based on a hybrid architecture. Implementing pixel-level operations in the FPGA and image-level operation using the embedded microprocessor allows parallel processing cutting down the execution time of the algorithms to meet real-time exigencies. The image processing module implemented is capable of finding the markers in the camera image and of extracting information relevant for navigation.
The task of the navigation module implementing visual servoing is to drive the robot by a path specified by the markers. By realizing a multi-level control architecture, the module uses feedback from the image processing module as well as the signals of incremental encoders of the wheels.
Appropriate operation of the application integrating these modules is validated by several tests.