Planning of multiple coexisting trajectories for autonomous vehicles based on self-organizing maps

In this work, we propose an algorithm for generating multiple simultaneous trajectories focusing on covering a given area for surveillance applications based on multiple Kohonen self-organizing maps. The proposed algorithm generates trajectories of optimal two-dimensional space filling, avoiding self and inter-trajectory crossings, allowing rapid update of the remaining trajectories if one or more vehicles fail. This strategy allows several levels of control in the properties of the trajectories, such as fixing or letting loose any of its points, changing the shape of regions after or during convergence, so providing a promising alternative for the generation of non-overlapping trajectories for multiple vehicles aimed at surveillance missions, for example, in the case of unmanned aerial vehicles (UAVs).

Automated summary

In this work, an algorithm based on multiple Kohonen self-organizing maps is proposed to generate multiple simultaneous trajectories for surveillance applications. The proposed algorithm generates trajectories that optimize two-dimensional space filling, avoid self and inter-trajectory crossings, and allow for rapid updates in case of vehicle failure. This strategy provides control over the properties of the trajectories, allowing for customization of point fixation, shape changes, and non-overlapping trajectories for multiple vehicles in surveillance missions, such as unmanned aerial vehicles (UAVs).