The thesis addresses the comparison between algorithms used in the field of local path planning for autonomous navigation of a mobile robotic system developed at the Laboratory of Robotics, Faculty of Electrical Engineering, University of Ljubljana. The algorithms tested during the research are: "Elastic band", "Time elastic band", "Nearness diagram", "Trajectory rollout" and "Dynamic window approach". The work involved the development of high-level logic to control a mobile robotic system (navigation framework) and the process of implementation and parameters tuning of the algorithms. Then, the evaluation principle of the algorithms was conceived and the corresponding scenarios were designed, which varied in complexity among each other. Once the design and development were completed, the platform was thoroughly tested in a real environment and all results were documented. The final objective was to evaluate all the algorithms on the basis of the results obtained from the ongoing testing, and thus to draw a conclusion as to which of the algorithms considered would be the most suitable for our mobile robotic system.
At the beginning of the diploma thesis, we have presented the reader with all the theoretical background necessary to understand how the platform works, as well as the knowledge regarding the navigation algorithms and the way in which the algorithms will be evaluated and compared with each other. The hardware of the platform, which is essential to establish good navigation in autonomous systems, was presented afterwards. We have presented the sensor equipment for navigation and the computer and industrial programmable logic controller (PLC) that are the basis for data processing. Then low- and high-level software were presented, with more emphasis on high-level software (ROS environment), as the thesis itself is more focused on things related to this equipment. This was followed by an explanation of the design of the scenarios and the design of the testing procedure.
In the last part, the results were presented and analyzed in order to compare the performance of the different algorithms. The results are presented for each area of comparison and for each scenario separately in order to make it as easy as possible for the reader to review the results. At the end of this section, we summarize the overall results and draw a final conclusion regarding the suitability of the algorithms for our mobile robotic platform. We found that the most suitable algorithms are "Elastic band" and "Time elastic band".
In the conclusion of this diploma thesis, we revisit the objectives and explain the overall workflow of the thesis. We also mention the conclusions we have drawn from the thesis. Finally, we outline possible upgrades and improvements to the research and describe possible future research directions in this area.
|
