Comparison of Trajectory Optimization Algorithms for High-Speed Quadrotor Flight Near Obstacles
Academic Article
Overview
Research
Identity
Additional Document Info
Other
View All
Overview
abstract
2016 IEEE. For autonomous quadrotors to be used for applications such as delivery, disaster response, and inspection, there is a need to fly near obstacles, especially in urban and indoor environments. Often, it is also beneficial to fly at high speeds to complete tasks quickly. A key challenge in enabling these capabilities is to plan trajectories, in a limited time period, that are safe, collision-free, and dynamically feasible. In this context, dynamic feasibility means that the trajectory can be accurately tracked in flight: a critical requirement when flying near to obstacles. When evaluating trajectory planning algorithms, it is important to understand their strengths and weaknesses; in this paper, we compare three state-of-the-art algorithms with a set of real-world planning scenarios and assessments of performance when planned trajectories are tracked in flight. We introduce a combination of the work of Campos-Macias et al. with Bry et al., which is shown to perform well for slow and conservative trajectories. The algorithm from Bry et al. is shown to perform the best for higher speed trajectories with relatively large amounts of open space. An adaptation of the algorithm from the work of Chamitoff et al. for quadrotor applications, as presented here, is shown to be the best planner for high-speed trajectories in tight confines, while relaxing the need for initial seed-paths that are collision free.
