Dynamics-aware navigation among moving obstacles with application to ground and flying robots
Spyridon G. Tarantos, Tommaso Belvedere, Giuseppe Oriolo · Robotics and Autonomous Systems · 2023
We present a novel method for navigation of mobile robots in challenging dynamic environments. The method, which is based on Nonlinear Model Predictive Control (NMPC), hinges upon a specially devised constraint for dynamics-aware collision avoidance. In particular, the constraint builds on the notion of avoidable collision state, taking into account the robot actuation capabilities in addition to the robot-obstacle relative distance and velocity. The proposed approach is applied to both ground and flying robots and tested in a variety of static and dynamic environments. Comparative simulations with an NMPC using a purely distance-based collision avoidance constraint confirm the superiority of the dynamics-aware version, especially for high-speed navigation among moving obstacles. Moreover, the results indicate that the method can work with relatively short prediction horizons and is therefore amenable to real-time implementation.