Development and Evaluation of Collision-Avoiding Sampling Strategies for Aerial Path Planning Algorithms

Dies ist die aktuellste Version dieses Eintrags.

Kurzfassung

This master thesis develops a new approach to collision avoidance in sampling algorithms in flight path planning, specifically for helicopters. The focus is on modifying sampled nodes whose direct connection to the search tree is blocked by obstacles in order to enable conflict-free integration. Four adaptive tactics are used for this purpose: (1) a climb angle adjustment for initially too steep connections, (2) a vertical shift to fly over obstacles, (3) a horizontal shift to avoid obstacles on the (x,y) plane, and (4) an adjustment of the heading, which adjusts the shape of the flight path and can also avoid obstacles. For experimental validation, the Open Motion Planning Library (OMPL) was adapted, in which a simplified helicopter motion model was implemented and the sampling methods RRT and RRT* were extended to include the displacement strategies described above. The benchmarks demonstrate that these modifications allow approximately 99% of the samples to be changed to a valid state and that, overall, shorter and smoother trajectories can often be achieved. At the same time, however, the results show a clear trade-off: the additional displacement steps increase the computational effort and resource consumption.

Verfügbare Versionen dieses Eintrags

• Development and Evaluation of Collision-Avoiding Sampling Strategies for Aerial Path Planning Algorithms. (deposited 28 Feb 2026 22:45) [Gegenwärtig angezeigt]