Document Type

Conference Proceeding

Publication Date

7-2018

Abstract

For self-driving mechanisms, the motion planning requires a reasonably fast algorithm for collision detection along the trajectories. We present three algorithms for the detection of collision among objects with predefined trajectories. The first algorithm uses the intersection of the path’s bounding box. The second algorithm sequentially checks for intersection between each pair of corresponding axis-aligned bounding boxes (AABB) from the trajectories of the two paths. Lastly, the latter algorithm is modified using iterative time advancement to an estimated earliest possible collision time. Simulation experiments on a variety of pair trajectories demonstrate a significant speedup of the proposed algorithms over the existing baseline algorithm. They are, therefore, preferable alternatives for faster broad-phase collision detection in applications such as motion planning.

Download

Share

COinS