| 引用本文: | 李昕哲,张波涛,汪江平,吕强,陈云.基于概率语义模型的最优期望时间目标搜索[J].控制理论与应用,2026,43(5):1043~1051.[点击复制] |
| LI Xin-zhe,ZHANG Bo-tao,WANG Jiang-ping,LU Qiang,CHEN Yun.Optimal expected-time based target search strategy with a probabilistic semantic model[J].Control Theory & Applications,2026,43(5):1043~1051.[点击复制] |
|
| 基于概率语义模型的最优期望时间目标搜索 |
| Optimal expected-time based target search strategy with a probabilistic semantic model |
| 摘要点击 315 全文点击 20 投稿时间：2024-01-06 修订日期：2025-11-19 |
| 查看全文 查看/发表评论 下载PDF阅读器 HTML |
| DOI编号 10.7641/CTA.2025.40013 |
| 2026,43(5):1043-1051 |
| 中文关键词 移动机器人 运动规划 概率语义模型 最优期望时间 |
| 英文关键词 mobile robot motion planning probabilistic semantic model optimal expected-time |
| 基金项目 国家自然科学基金项目(62073108, U22A2044), 浙江省自然科学基金项目(LZ23F030004)资助. |
|
| 中文摘要 |
| 针对目前离散目标搜索方法不适于在3D动态环境中进行目标搜索的问题, 本文提出了一种基于概率语义
地图的3D概率离散双层规划策略(3D-PDBP), 所述方法可用于搜索位置不确定的低机动性目标, 适用于目标短期内
不会连续变化的场景. 为实现目标分布概率的动态更新, 基于艾宾浩斯遗忘曲线构建了仿人记忆与遗忘机制的概
率动态更新模型HMF-PU. 结合目标的先验语义信息和分布概率, 基于包含环境观测点的拓扑地图, 构建了适用于
目标搜索的概率语义地图. 3D-PDBP将观测点搜索序列规划与深度视觉传感器运动规划相结合, 基于概率语义地
图规划观测点搜索序列, 并根据目标在3D空间中的概率分布模型规划视觉传感器的伺服过程. 实验结果表明3DPDBP
可在不确定环境中高效完成目标搜索任务, HMF-PU可有效平衡目标信息的重要性与实时性. |
| 英文摘要 |
| This paper proposes a novel method for 3D probabilistic discrete bi-level programming (3D-PDBP) to address
the limitations of current discrete target search methods in dynamic 3D environments. The proposed method enables
the search for low-maneuverability targets with uncertain positions, particularly suited for scenarios where the target’s
state changes intermittently in the short term. To dynamically update the distribution probability of the target, the paper
introduces a probabilistic dynamic update model HMF-PU that can mimic human memory and a forgetting mechanism
is built based on the Ebbinghaus forgetting curve. By combining the target’s prior semantic information with distribution
probability, the paper constructs a probabilistic semantic map for effective target search. The proposed 3D-PDBP combines
observation point search sequence planning with depth vision sensor motion planning, planning the observation point search
sequence based on the probabilistic semantic map and orchestrating the servo process of the visual sensor using the target’s
probability distribution model HMF-PU. Experimental results reveal that 3D-PDBP can accomplish target search tasks
efficiently in unpredictable situations, and HMF-PU can successfully balance the importance and real-time performance of
target information. |
|
|
|
|
|