Answers are grounded in this paper and cite their sources.
Agricultural harvesting manipulators are key components of smart agricultural harvesting systems, where collision-free and low-damage motions must be generated under irregular crop distribution, foliage occlusion, illumination variation, deformable obstacles, and uncertain target or obstacle boundaries. In such environments, path planning is influenced not only by geometric feasibility, but also by visibility, sensing reliability, contact safety, and execution stability. This review surveys path planning studies for agricultural harvesting manipulators from the perspective of optimization objectives and methodological development. Unlike crop-specific harvesting robot reviews or broad agricultural robot path-planning surveys, this paper focuses on manipulator-level motion generation across harvesting scenarios. Classical optimization-based methods are first summarized, including single-objective formulations focused on time, path length, energy consumption, smoothness, or safety, as well as multi-objective approaches that address trade-offs among competing criteria. These methods provide the main analytical foundation for agricultural manipulator path planning, but their performance often depends on manually designed objectives, weight settings, and relatively explicit environmental representations, limiting adaptability under changing field conditions. The review then discusses the growing use of learning-based methods, with particular attention to deep reinforcement learning and active learning. These methods are considered mainly as extensions to classical planning, especially for improving adaptability, sample efficiency, and deployment under uncertainty, rather than as direct replacements for model-based approaches. Finally, open challenges are discussed, including contact-aware manipulation, uncertainty-aware planning, task-level robustness, perception-planning-control coupling, and reproducible evaluation for smart agricultural harvesting systems. This review aims to clarify how optimization objectives, environmental uncertainty, and learning-based adaptation can support more robust and deployable path planning for agricultural harvesting manipulators.
No takes yet. Share an insight, caveat, or question.
Zhao et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: