Robust Sequential Convex Programming for Quadrotor Trajectory Planning with Obstacle Avoidance Under Wind Disturbances

Abstract

In this work, we address the problem of trajectory planning for a 3D quadrotor navigatingaround three spherical obstacles under bounded wind disturbances. The goal is to reach a desired final state from a given initial state within a fixed time, while minimizing fuel consumption and avoiding obstacles. We first compute a nominal trajectory in a no-wind setting using Sequential Convex Programming (SCP) applied to a 6-state, 3-DoF quadrotor model. This trajectory satisfies all state, control, and obstacle constraints while minimizing fuel use. However, in the presence of constant wind, the nominal path becomes unreliable due to unmodeled drift, leading to potential obstacle violations—unacceptable in safety-critical applications. To improve robustness without relying on any low-level controller, we implement three strategies: (i) LQR-based tracking of the wind-free nominal trajectory, (ii) a receding-horizon SCP approach that re-solves the optimization at each node using updated state feedback, convex half-space approximations, and Wind-Adaptive Residual Correction (WARC), and (iii) a tracking-style method using smaller SCP subproblems to increase responsiveness. Finally, we apply funnel synthesis along the nominal trajectory to certify allowable deviations at each node. This framework enables safe, fuel-efficient flight despite bounded disturbances.