.png){kind=icon}
.png){kind=icon}
Return
Waypoint Kinodynamic Motion Planning for a Tractor with Two Trailers in an Orchard: Sampling-Based vs. Optimization- Based Approaches
.png){kind=icon}
OA
AI
M
S
DOI：10.3390/agriculture16141490.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
We propose a novel sampling-based approach to waypoint-constrained kinodynamic planning for a custom apple-picking machine consisting of a tractor towing two trailers, operating autonomously in an orchard. The platform must follow sparse 2D waypoints through narrow headlands and rows. We adapt the Asymptotically Optimal A* (AOA*) planner to this setting and compare it against optimization-based methods (Fatrop and Ipopt) on representative benchmarks, a headland U-Turn, complex long path, and an environment with random obstacles. AOA* finds first feasible plans in seconds (often sub-second for U-turns), achieves 100% success at moderate temporal resolution, and reduces trajectory-time suboptimality to under 7% with a short anytime budget. Optimization-based planners (Fatrop and Ipopt) reach near-optimal trajectories when successful but typically require coarser time steps and longer runtime. At the same time, sampling-based algorithms support arbitrary obstacle geometries and are easier to set up. Overall, AOA* finds a feasible plan faster and scales better with the horizon and temporal resolution. At the same time, optimization-based solvers produce smoother, near-locally optimal trajectories when warm-started. The results of this work can enable a generalized approach for both intra-row maneuvers and inter-field logistics (e.g., orchard–warehouse transits) without changing environment representations.
Keywords:
tractor–trailer planning
kinodynamic planning
sampling-based planning
optimization-based planning
Journal
IF:
3.6
Papers:
1.3W
Citations:
2.8W
Organization
Cited Papers