Gradient Descent Algorithm for Quadrotor Safe Path Planning in Complex Environments

In this paper, we propose an algorithmic framework for local path planning using gradient descent in complex environments, where we divide the trajectory planning problem into two aspects: global path search and local path planning. We use Jump Point Search (JPS) as the global path planning algorithm to find a path that safely crosses an obstacle. In local path planning, we introduce a gradient descent algorithm to optimise the local trajectory segments by minimising the distance, curvature and safety distance to generate high-quality flight trajectories. Simulation results show that our proposed improved algorithm exhibits higher smoothness, stability and safety in complex environments compared to the traditional algorithm. In the simulation environment, the improved algorithm can significantly shorten the actual flight time, reduce the total path length, and lower the energy consumption, thus improving the smoothness and efficiency of the flight.