High-precision Autonomous Flight and Soft-Landing of Multirotor UAVs with Terminal Attitude Constraints

Abstract

With the rapid expansion of drone applications, UAVs are increasingly expected to operate in complex and dynamic environments such as landing on autonomous ground vehicles, moving ships, or mobile delivery platforms which introduces significant challenges in trajectory generation, demanding precise control over both position and orientation during touchdown to prevent, slippage, or toppling. This thesis addresses the problem of high-precision flight and soft-landing of multirotor UAVs under such conditions, where limited onboard computational resources and external disturbances further complicate autonomous operation.

First, the problem of soft-landing on a static platform is addressed using a minimum-jerk-based guidance strategy that enforces ‘hard constraints’ on position, velocity, and acceleration, with a continuously updated optimal time-to-go making the system robust to path perturbations. To further improve energy efficiency, the guidance formulation is extended to minimize both jerk and acceleration. A semi-analytical method to compute time-to-go and associated weighting parameter is developed. The framework is adapted to land on moving platforms by incorporating relative dynamics and time-varying terminal conditions. By regulating the terminal acceleration, which is intrinsically coupled with attitude, the strategy is extended to facilitate landing on oscillating platforms at a desired attitude, necessitating the prediction of the platform’s orientation. Finally, by combining relative dynamics with continuously updated terminal boundary conditions derived from predicted platform attitudes, the challenge of landing on moving and oscillating platforms is effectively addressed.

These methods are extensively validated through numerical and software-in-the-loop (SITL) simulations and implemented within a guidance, navigation, and control (GNC) module onboard a quadrotor UAV, with the complete system experimentally validated through numerous real-world flight trials demonstrating successful autonomous soft-landing on both static and dynamic platforms.