面向低光弱纹理环境的SLAM系统实验教学平台设计
Design of Experimental Teaching Platform for SLAM System in Low-light and Weak-texture Environment
叶涛 1赵云龙 1汪寿安 1李允旺 1严翔明1
作者信息
- 1. 中国矿业大学(北京)机械与电气工程学院,北京 100083
- 折叠
摘要
为提升机器人专业实验教学质量和人才培养质量,针对机器人状态估计和精确定位工程问题提出一种面向低光弱纹理环境的视觉惯性融合鲁棒SLAM算法,并设计一种面向低光弱纹理环境的SLAM系统实验软件教学平台,实现稳健鲁棒4DOF状态估计和高精度实时定位,并在公开数据集上验证了该算法的有效性.实验结果表明,与现有的视觉惯性SLAM算法相比,所提算法在结构化弱纹理场景中有着更高的精度和鲁棒性,能有效降低机器人系统的累积误差,保证地图构建的一致性.通过上述SLAM系统教学平台在实验教学中的应用,可提升机器人专业本科生在工程问题中的创新能力和实践能力.
Abstract
In order to improve the quality of professional robot experimental teaching and the quality of talent cultivation,this paper proposes a robust visual-inertial fusion SLAM algorithm tailored for low-light and weak-texture environment,and designs a software-based experimental teaching platform for SLAM system in the environment.The proposed approach achieves robust 4DOF state estimation and high-precision real-time localization,and its effectiveness is validated through experiments on publicly available datasets.The experimental results demonstrate that compared to existing visual-inertial SLAM algorithms,the proposed algorithm exhibits higher accuracy and robustness in structured weak-texture scenes,effectively reduces the cumulative errors of the robot system and ensuring map consistency.By utilizing the proposed SLAM system teaching platform in experimental instruction,the innovative and practical capabilities of undergraduate students majoring in robotics are enhanced when dealing with real-world engineering challenges.
关键词
弱纹理/视觉惯性导航/同步定位与建图/实验教学Key words
weak texture/vision IMU navigation/SLAM/experimental teaching引用本文复制引用
基金项目
国家自然科学基金项目(52374166)
中国矿业大学(北京)"课程思政"示范课程建设项目(SZ230502)
课程思政项目(YKCSZ2023013)
中国矿业大学(北京)本科教育教学改革项目(J230416)
出版年
2024
NETL
NSTL
维普
万方数据