Purpose

With the booming development of computer, optical and sensing technologies and cybernetics, the technical research in unmanned vehicle has been advanced to a new era. This trend arouses great interest in simultaneous localization and mapping (SLAM). Especially, light detection and ranging (Lidar)-based SLAM system has the characteristics of high measuring accuracy and insensitivity to illumination conditions, which has been widely used in industry. However, SLAM has some intractable problems, including degradation under less structured or uncontrived environment. To solve this problem, this paper aims to propose an adaptive scheme with dynamic threshold to mitigate degradation.

Design/methodology/approach

We propose an adaptive strategy with a dynamic module is proposed to overcome degradation of point cloud. Besides, a distortion correction process is presented in the local map to reduce the impact of noise in the iterative optimization process. Our solution ensures adaptability to environmental changes.

Findings

Experimental results on both public data set and field tests demonstrated that the algorithm is robust and self-adaptive, which achieved higher localization accuracy and lower mapping error compared with existing methods.

Originality/value

Unlike other popular algorithms, we do not rely on multi-sensor fusion to improve the localization accuracy. Instead, the pure Lidar-based method with dynamic threshold and distortion correction module indeed improved the accuracy and robustness in localization results.

中文翻译：

---

基于激光雷达的SLAM退化抑制自适应方案

目的

随着计算机、光学、传感技术和控制论的蓬勃发展，无人车技术研究进入了一个新的时代。这种趋势引起了人们对同步定位和映射（SLAM）的极大兴趣。尤其是基于光探测与测距（Lidar）的SLAM系统具有测量精度高、对光照条件不敏感等特点，在工业上得到了广泛的应用。然而，SLAM 有一些棘手的问题，包括在结构性较差或人为的环境下的退化。为了解决这个问题，本文旨在提出一种具有动态阈值的自适应方案来减轻退化。

设计/方法/方法

我们提出了一种具有动态模块的自适应策略，以克服点云的退化。此外，局部地图中引入了失真校正过程，以减少迭代优化过程中噪声的影响。我们的解决方案确保对环境变化的适应性。

发现

在公开数据集和现场测试的实验结果表明，该算法具有鲁棒性和自适应性，与现有方法相比，实现了更高的定位精度和更低的映射误差。

原创性/价值

与其他流行算法不同，我们不依赖多传感器融合来提高定位精度。相反，具有动态阈值和失真校正模块的纯基于激光雷达的方法确实提高了定位结果的准确性和鲁棒性。