Autonomous Planning, Navigation and Control for Lightweight Unmanned Aerial Vehicles in Cluttered Environments

This paper proposed an autonomous planning, navigation and control framework for lightweight unmanned aerial vehicles (UAVs) in obstacle-dense environments with limited computational resources. The framework employed a polynomial-based minimum-snap trajectory generation method to ensure obstacle-free in confined three-dimensional (3D) spaces. A gradient-based trajectory optimization strategy integrated smoothness constraints, collision avoidance and dynamic feasibility was proposed to ensure safety and motion smoothness. The model predictive control (MPC) method was utilized to enhance the trajectory tracking ability of UAVs by reconciling the optimized reference path with real-time sensory inputs. Simulations and outdoor flight experiments demonstrated the effectiveness of the proposed framework.