A safety-refined and smoothness-enhanced path-planning algorithm for an agricultural composite mobile manipulator in greenhouse crate handling

With the growing demand for automation in greenhouse logistics, ensuring both operational safety and motion smoothness has become a key challenge for composite mobile manipulators working in confined agricultural environments. This study proposes a safety-refined and smoothness-enhanced path-planning algorithm, termed SR-RRT-APF, to improve path feasibility and collision avoidance for agricultural robotic systems. The method integrates scheduled goal biasing, curvature-aware parent-node selection, density-adaptive step sizing, and potential-field-based soft guidance into an improved RRT framework. By incorporating explicit minimum-clearance constraints and lightweight post-processing, the algorithm jointly optimizes safety margins and geometric smoothness during the path generation stage. Extensive simulations and prototype-level tests were conducted on a greenhouse crate-handling robot equipped with a 6-DOF manipulator and a vision-guided mobile chassis. Ten consecutive crate-handling cycles were performed, in which the robot autonomously recognized, grasped, transported, and placed vegetable crates within narrow greenhouse aisles. Results from the simulation benchmarks show that SR-RRT-APF achieves superior path quality, larger safety margins, and improved smoothness compared with the benchmark algorithms in dense and constrained workspaces. Prototype-level experiments on a greenhouse crate-handling robot further support the system-level feasibility of the associated perception–manipulation workflow, indicating the practical relevance of the proposed method in greenhouse operations while also suggesting its applicability to a broader class of constrained-space planning problems.