A Perception-Aware NMPC for Vision-Based Target Tracking and Collision Avoidance with a Multi-Rotor UAV

TL;DR

This paper introduces a perception-aware NMPC strategy for multi-rotor UAVs that enables vision-based target tracking and collision avoidance while respecting actuation and perception constraints.

Contribution

It presents a novel NMPC approach that integrates perception constraints with collision avoidance for diverse multi-rotor platforms in dynamic environments.

Findings

Successful simulation validation in MATLAB

Effective target tracking with perception constraints

Safe navigation among dynamic obstacles

Abstract

A perception-aware Nonlinear Model Predictive Control (NMPC) strategy aimed at performing vision-based target tracking and collision avoidance with a multi-rotor aerial vehicle is presented in this paper. The proposed control strategy considers both realistic actuation limits at the torque level and visual perception constraints to enforce the visibility coverage of a target while complying with the mission objectives. Furthermore, the approach allows to safely navigate in a workspace area populated by dynamic obstacles with a ballistic motion. The formulation is meant to be generic and set upon a large class of multi-rotor vehicles that covers both coplanar designs like quadrotors as well as fully-actuated platforms with tilted propellers. The feasibility and effectiveness of the control strategy are demonstrated via closed-loop simulations achieved in MATLAB.