Application-oriented Path Following Control for Unmanned Aerial Vehicles

Abstract

Controlling Unmanned Aerial Vehicles (UAVs) in obstacle-rich environments is challenging and typically requires a hierarchical chain of control algorithms, dividing the problem for computational feasibility. For this separation to work properly, a precise and robust path following control algorithm is necessary. UAVs exhibit highly nonlinear dynamics that complicate precise path following for control algorithms on today's computationally limited flight control hardware. This motivates the search for efficient control algorithms with a focus on real-time applicability. We compare the advantages and disadvantages of path following and trajectory tracking algorithms when combined with the rest of the hierarchical control chain. Two approaches are explored in this work. An extension of the integrated Proportional Integral Derivative (PID) control pipeline of the PX4 Autopilot platform and a sampling-based Model Predictive Control (MPC) approach using the Model Predictive Path Integral Control (MPPI) control framework. For both approaches, we develop a path following and a trajectory tracking version. A novel checkpoint-based geometric control strategy is derived for the path following version of the PID control pipeline extension. We validate the developed control algorithms through simulation in Gazebo and real-world flight tests. The results demonstrate the real-time capability of both algorithms on current flight control hardware. Both approaches reveal inherent differences when comparing their performance. The proactive nature of the MPPI algorithm contrasts the knowledge of the past in PID algorithms.