A Two Step Velocity Planning Method for Real-time Collision Avoidance of Multiple Aerial Robots in Dynamic Environments

摘要

This paper presents a collision avoidance method to improve the safety conditions in scenarios with multiple UAVs sharing the same aerial space with other aircrafts or mobile obstacles. The method modifies the velocity profile of the UAVs under control maintaining the paths initially planned. It is based on the combination of both a Search Tree algorithm which finds a solution if it exists, and the minimization of a cost function which tries to find the nearest solution to the initially planned trajectories for the UAVs. The Search Tree algorithm provides an initial valid order of pass for the vehicles involved in a given conflict and allows to formulate the minimization problem as a LP problem that can be efficiently solved. A model for the UAVs has been considered in the paper and several simulations are presented with promising results.