Vývoj samořiditelné platformy

Abstract

This thesis is focused on the development of software for a self-driving platform, which is being developed by the Technical University of Ostrava. Upon completion, the software should be capable of safely navigating an autonomous vehicle through the university campus, taking inputs from image analysis software, GPS, LiDAR, and the vehicle's own controller unit.

The core of the path planning software is split into two components. The first component handles communication with the vehicle's controller unit via a CAN bus. The second component then does all the path planning, taking into account the aforementioned inputs. The second component then asks the first component to maintain a certain speed or steer the vehicles, however, it doesn't have to handle CAN messages, including driving inputs or control checksums, by itself. The two components communicate using the Transmission Control Protocol. The path planning software can be tested using a simulator, capable of mocking the CAN handling software, along with the entire vehicle. The simulator honors the TCP API, however, the vehicle physics are simplified. PID controllers are used in multiple parts of the software.

At the time of publishing this paper, the software is still unfinished, as the self-driving platform is an academic research project, which outspans this bachelor's thesis in scope, and the development of the platform is set to continue.