The work deals with the problem of planning a path of a mobile robot in an environment with obstacles. Special emphasis is placed on the tangent beetle algorithm, which, unlike global methods, does not require a prior map, but is based on local sensor readings, such as LIDAR. The aim of the thesis was to implement the algorithm, test it empirically, and compare it with other established algorithms, such as A*, Dijkstra, and algorithms from the beetle family.
As part of the task, a software package was developed in Python that enables the generation of random test environments, the execution of simulations, and the visualization of results. The influence of various environmental parameters, such as the number and size of obstacles and the range of the LIDAR sensor, on the path length and time complexity of the algorithms was analyzed. The results of the experiments show that the tangent beetle algorithm achieves shorter paths on average than other local algorithms, but has a significantly higher time complexity, which increases exponentially with the complexity of the environment.
The task confirms theoretical expectations and shows that the tangent beetle is suitable for use in scenarios where path optimality is key, but less effective with a larger number of obstacles or a longer sensor range, where its scalability proves to be limiting.
|
