A Multi-task-oriented Path Planning Framework for Agricultural Robots Based on A ^∗ Hybrid Firefly Algorithm

Agricultural robots play a pivotal role in enhancing the efficiency and intelligence of agricultural operations. However, most existing path planning methods focus primarily on point-to-point navigation or single-task execution, making them inadequate for the coordinated performance of multiple tasks in complex farmland environments. To address this challenge, this paper proposes a task-oriented path planning framework that integrates the A ∗ algorithm with an improved Firefly Algorithm (FA). The framework first formulates a Traveling Salesman Problem (TSP) model that concurrently considers warehouse constraints and obstacle avoidance. A 16-directional A ∗ algorithm is employed to achieve efficient and collision-free path search. To enhance path guidance and search efficiency, a directional angle penalty factor and a distance expansion factor are incorporated into the heuristic function. During the task scheduling optimization stage, a novel sequencing encoding mechanism is designed, combined with Lévy flight, adaptive mutation, and elite retention strategies, significantly improving the global search capability and convergence stability of the FA in multi-task path optimization. Extensive simulations conducted on various representative farmland maps, including both single- and multi-target scenarios, demonstrate that the proposed method outperforms mainstream algorithms in terms of path smoothness, planning efficiency, and scheduling performance. This work provides an effective technical foundation and theoretical support for intelligent operations of agricultural robots in unstructured terrains.