用强化学习统一解决草莓采摘中的避障、摘果和放置问题。
Reinforcement Learning for the Full Strawberry Harvesting Process: Obstacle Separation, Detachment, and Placement
- 用统一策略处理采摘全流程，结合启发式逻辑协调任务进度。
- 真实场景下采摘成功率达82.0%，随遮挡程度增加执行时间延长。
- 适合研究机器人采摘、复杂交互控制的科研人员参考。
严重遮挡和可变形植株结构带来复杂的接触动力学，挑战机器人草莓采摘。本文提出一种以策略驱动的强化学习框架，将避障、果实分离与放置整合为序列决策任务。共享的交互感知策略生成各阶段笛卡尔运动，轻量级启发式逻辑协调任务推进与夹爪动作。采用共享结构化观测空间表示目标、障碍物、末端执行器及任务上下文信息。分层架构结合高层策略与低层笛卡尔阻抗控制，实现柔顺交互。为支持零样本仿真到现实迁移，采用可行性优先观测对齐与领域随机化。策略在仿真中成功率达89.7%，真实实验中达82.0%。当遮挡等级从1增至5时，平均执行时间由12.99秒升至21.73秒，反映交互复杂性提升。结果证明了交互感知采摘行为可有效迁移至结构不同的机器人平台。
原文摘要 · Abstract (English)
Severe occlusions and deformable plant structures introduce complex contact dynamics that challenge robotic strawberry harvesting. A policy-driven reinforcement learning (RL) framework with heuristic phase coordination was developed, in which obstacle separation, fruit detachment, and placement were formulated as a sequential decision-making task. A shared interaction-aware policy generated Cartesian motions across all task phases, while lightweight heuristic logic coordinated task progression and gripper events. A shared structured observation space was used to represent target, obstacle, end-effector, and task-context information. A hierarchical architecture combined the high-level policy with low-level Cartesian impedance control for compliant interaction. To support zero-shot sim-to-real transfer, feasibility-first observation alignment and domain randomization were adopted. The policy achieved success rates of 89.7% in simulation and 82.0% in real-world experiments. As the occlusion level increased from 1 to 5, the average execution time increased from 12.99 s to 21.73 s, reflecting greater interaction complexity. These results demonstrated effective transfer of interaction-aware harvesting behaviors to a structurally different robotic platform.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。