基于CBAM-DynaSLAM算法的居家照护机器人视觉感知优化方法研究

关键词：视觉感知  居家照护机器人  动态SLAM  注意力机制  ORB特征提取

Keywords:visual perception  home care robot  dynamic SLAM  attention mechanism  ORB feature extraction

摘要：

      摘要 目的：本文提出一种基于CBAM-DynaSLAM算法的居家照护机器人视觉感知优化方法，旨在提升居家照护机器人在动态环境下的目标感知精度，为其在居家照护场景中的实际应用提供技术支持。 方法：该方法在DynaSLAM框架的ORB（Oriented FAST and Rotated BRIEF）特征提取器中集成了卷积块注意力模块，通过增强特征提取的准确性，显著提升了动态居家照护场景下目标识别与跟踪的性能。 结果：相较于原始DynaSLAM，CBAM-DynaSLAM系统的绝对轨迹误差（ATE）降低了65%，相对姿态误差（RPE）降低了20%，验证了CBAM机制在提升动态场景鲁棒性与定位精度方面的有效性。 结论：本文提出的方法通过提升特征提取的准确性，显著提高了居家照护机器人在动态环境下的目标感知精度，有助于推动居家照护机器人在实际居家照护动态场景下的应用。

Abstract:

      Abstract Objective： This paper proposes a visual perception optimization method for home care robots based on the CBAM-DynaSLAM approach. It aims to enhance the target perception accuracy of home care robots in dynamic environments， providing technical support for their practical application in home care scenarios. Method： Current method integrates a convolutional block attention module in the ORB（Oriented FAST and Rotated BRIEF） feature extractor of the DynaSLAM framework，which significantly improves the performance of target recognition and tracking in dynamic homecare scenarios by enhancing the accuracy of feature extraction. Result： The experimental results show that compared with the original DynaSLAM，the CBAM-DynaSLAM system reduces the absolute trajectory error（ATE） by 65% and the relative pose error（RPE） by 20%，which verifies the effectiveness of the CBAM mechanism in improving the robustness and localization accuracy in dynamic scenes. Conclusion： The method proposed in this paper significantly improves the target perception accuracy of home care robots in dynamic environments by enhancing the accuracy of feature extraction， which helps to promote the application of home care robots in practical dynamic home care scenarios.