Efficient Unmanned Aerial Vehicle Path Planning In Complex Environments

Résumé

Unmanned Aerial Vehicles (UAVs) face significant challenges when navigating complex three-dimensional environments, particularly in the presence of irregularly shaped obstacles and dynamic changes. This thesis presents a fast and computationally efficient path planning framework based on an enhanced version of the Tangent Intersection Guidance algorithm, referred to as TIG*, capable of generating smooth and collision-free paths in static, partially known, and completely unknown environments. The proposed algorithm operates in three-dimensional environments where complex obstacles are represented as vertical prisms derived from their two-dimensional convex hulls. Feasible waypoints are generated by identifying the first intersected obstacle and computing corresponding tangent-based guidance points. To reduce computational complexity and improve path optimality, an A*-based mechanism is integrated to efficiently guide the search toward obstacle-associated waypoints. In addition, a local quadratic Bézier-based smoothing technique is applied to refine sharp turns along the planned path, resulting in improved path continuity while preserving obstacle clearance. The performance of TIG* is extensively evaluated through simulation in a wide range of scenarios, including variations in environment size, obstacle density, and altitude. The algorithm is compared against several widely used path planners, including PRM*, RRT*, Informed RRT*, APF, RHRRT*, and 3D-TG. The results demonstrate that TIG* achieves up to 30% shorter paths, significantly improved smoothness in terms of reduced turning radius, and substantially lower computation times, typically below 0.05 seconds across a wide range of evaluated scenarios, which is suitable for real-time UAV applications. Furthermore, the generated paths are validated through trajectory tracking using a linear Model Predictive Control (LMPC) framework applied to a quadrotor UAV. The tracking results confirm that the smoothed TIG* trajectories can be accurately followed with minimal position and attitude errors. Overall, the proposed TIG* framework provides a robust, efficient, and practical solution for real-time UAV path planning and execution in complex three-dimensional environments.

Citer ce document

Accès au document

Ce document est hébergé sur son dépôt institutionnel d'origine.

Auteur(s)

Statistiques

Consultations : 20

Téléchargements : 0