Local planner selection for autonomous robots using machine learning

Autonomous robots have been actively involved in visible and invisible aspects of our daily lives for several years. These robots can perform various tasks and roles in different locations, such as functioning as waiters in restaurants, serving as housekeepers in hotels, or transporting goods as attendants in warehouses. Autonomous robots rely on local planners to navigate in their environment. Nevertheless, selecting the most appropriate local planner from the available options for a spesific task can be challenging. Despite extensive research in areas such as autonomous navigation, human-robot interaction, and robot localization, there has been no study to determine the optimal local planner for the given environments. In this study, a machine learning solution that identifies the most suitable local planner from the available options prior to initiating navigation. Our approach involves conducting tests with different local planners under identical conditions in static environments, subsequently recording the data from each test. The recorded data is subsequently employed in the generation of a dataset for the purposes of training and testing the model. A total of 1920 tests were conducted in a simulation environment to create this dataset, which was then divided into training and test sets, with 80% allocated for training and 20% reserved for testing. The results demonstrate that the proposed solution achieves a success rate of up to 91% has been achieved in recommending the appropriate local planners based on the desired criteria.