Efficient and Safe Trajectory Planning for Autonomous Agricultural Vehicle Headland Turning in Cluttered Orchard Environments

TL;DR

The paper tackles safe and efficient headland turning for autonomous agricultural vehicles in cluttered orchards where headland space is limited. It introduces a two-stage planner: a front-end enhanced hybrid A* for rapid, collision-aware trajectory initialization and a back-end MINCO-inspired nonlinear optimizer that refines the trajectory under kinematic and collision constraints, using a differential-flatness representation with flat outputs $[\sigma_x,\sigma_y]^T$ and a state relation $[x,y,v,\theta,\dot{\theta},a,\kappa]$. To reduce conservatism, the vehicle and its implements are modeled as multiple rectangles with separate convex safe corridors, and collision checks are accelerated via a covering-circle inflation technique. The method achieves substantial computation-time reductions (e.g., front-end ~$84.6\%$, back-end ~$91.0\%$) and demonstrates real-time viability in both simulation and field trials on a vineyard robot, indicating strong potential for broader AAV deployment in complex orchards.

Abstract

Autonomous agricultural vehicles (AAVs), including field robots and autonomous tractors, are becoming essential in modern farming by improving efficiency and reducing labor costs. A critical task in AAV operations is headland turning between crop rows. This task is challenging in orchards with limited headland space, irregular boundaries, operational constraints, and static obstacles. While traditional trajectory planning methods work well in arable farming, they often fail in cluttered orchard environments. This letter presents a novel trajectory planner that enhances the safety and efficiency of AAV headland maneuvers, leveraging advancements in autonomous driving. Our approach includes an efficient front-end algorithm and a high-performance back-end optimization. Applied to vehicles with various implements, it outperforms state-of-the-art methods in both standard and challenging orchard fields. This work bridges agricultural and autonomous driving technologies, facilitating a broader adoption of AAVs in complex orchards.

Figures (8)

• Figure 1: Illustration of an electric tractor equipped with a pruner executing a turning trajectory in a headland.
• Figure 2: Vineyard tractors with: (a) double-sided pruner, (b) single-sided pruner, (c) mower, and (d) KMS sprayer, along with their geometric representations for trajectory planning.
• Figure 3: Process of determining optimal covering circles for vehicle and implement. The black areas represent trees, and the gray areas represent map inflation.
• Figure 4: Illustration of (a) geometric representations of obstacles and the AAV; (b) results from the front-end search using covering circles; (c) the parameterized initial trajectory with multiple segments, piece points, and constraint points; and (d) the optimized trajectory following the back-end refinement.
• Figure 5: Comparison between (a) our planner and (b) SRA results carrying a KMS sprayer in Case II. The red dashed line shows the front-end trajectory and the green line shows the optimized trajectory from the back end. Safe corridors for both the vehicle and implement are depicted. The covering rectangle in SRA is shown with dashed black lines, and the dashed blue lines represent the pull-out trajectories.