Lidar SLAM Algorithm Based on Point Cloud Geometry and Intensity Information

Long Peng; Weigang Li; Qifeng Wang; Huan Yi

doi:10.3788/LOP241834

Journals >Laser & Optoelectronics Progress >Volume 62 >Issue 8 >Page 0815001 > Article

Laser & Optoelectronics Progress
Vol. 62, Issue 8, 0815001 (2025)

Lidar SLAM Algorithm Based on Point Cloud Geometry and Intensity Information

Long Peng^1,*, Weigang Li¹, Qifeng Wang¹, and Huan Yi²

Author Affiliations

¹School of Information Science and Engineering, Wuhan University of Science and Technology, Wuhan 430081, Hubei , China

²Hubei New Industry Technician Institute, Xianning 437000, Hubei , China

show less

DOI: 10.3788/LOP241834 Cite this Article Set citation alerts

Long Peng, Weigang Li, Qifeng Wang, Huan Yi. Lidar SLAM Algorithm Based on Point Cloud Geometry and Intensity Information[J]. Laser & Optoelectronics Progress, 2025, 62(8): 0815001 Copy Citation Text

show less

Abstract

To address the low accuracy of the laser simultaneous localization and mapping (SLAM) algorithm in large scenes, this paper proposes a lidar SLAM algorithm based on point cloud geometry and intensity information. First, point cloud geometry and intensity information are combined for feature extraction, where weighted intensity is introduced in the calculation of local smoothness to enhance the robustness of feature extraction. Subsequently, a rectangular intensity map is introduced to construct intensity residuals, whereas geometric and strength residuals are integrated to optimize pose estimation, thereby improving mapping accuracy. Finally, a feature descriptor based on point cloud geometry and intensity information is introduced in the loop detection process, thus effectively enhancing the loop recognition accuracy. Experimental results on public datasets show that, compared with the current mainstream LeGO-LOAM algorithm, the proposed algorithm offers greater position accuracy.

Keywords

feature descriptor intensity information LiDAR loop detection simultaneous localization and mapping

P_{r} = \frac{P_{e} D_{r}^{2} ρ}{4 R^{2}} η_{s y s} η_{a t m} c o s α

（1）

I = \frac{P_{r}}{P_{e}} = \frac{D_{r}^{2} ρ}{4 R^{2}} η_{s y s} η_{a t m} c o s α

（2）

ρ \propto \frac{I R^{2}}{c o s α}

（3）

n = \frac{(p - p_{1}) \times (p - p_{2})}{|p - p_{1}| \cdot |p - p_{2}|}

（4）

c o s α = \frac{p^{T} \cdot n}{|p|}

（5）

c_{G i} = \frac{1}{| S |} \sum_{j \in S, j \neq i} (|P_{i} - P_{j}|)

（6）

c_{I i} = \frac{1}{| S |} \sum_{j \in S, j \neq i} (|I_{i} - I_{j}|)

（7）

c = w_{G i} c_{G i} + w_{I i} c_{I i}

（8）

w_{G i} = e x p (- |G_{i} - \bar{G}|)

（9）

w_{I i} = e x p (- |I_{i} - \bar{I}|)

（10）

f_{e} (p_{i}) = p_{n} \cdot [(T_{k} p_{i} - p_{i l}) \times n_{i l}]

（11）

p_{n} = \frac{(T_{k} p_{i} - p_{i l}) \times n_{i l}}{‖(T_{k} p_{i} - p_{i l}) \times n_{i l}‖}

（12）

f_{s} (p_{i}) = (T_{k} p_{i} - p_{i s}) \cdot n_{i s}

（13）

I_{k} = \frac{1}{n} \sum_{m \in M, m \neq 0} (|I_{m} - I_{n e a r}|) (k \in [1,8])

（14）

I_{12} = \frac{x_{2} - x_{i}}{x_{2} - x_{1}} I_{1} + \frac{x_{i} - x_{1}}{x_{2} - x_{1}} I_{2}

（15）

I_{L} = \frac{y_{3} - y_{i}}{y_{3} - y_{2}} I_{12} + \frac{y_{i} - y_{2}}{y_{3} - y_{2}} I_{34}

（16）

I_{H} = \frac{y_{7} - y_{i}}{y_{7} - y_{6}} I_{56} + \frac{y_{i} - y_{6}}{y_{7} - y_{6}} I_{78}

（17）

{\hat{I}}_{p_{i}} = \frac{z_{5} - z_{i}}{z_{5} - z_{1}} I_{L} + \frac{z_{i} - z_{1}}{z_{5} - z_{1}} I_{H}

（18）

f_{I} (p_{i}) = {\hat{I}}_{p_{i}} - (I_{i} - I_{n e a r})

（19）

T_{k}^{*} = \underset{T_{k}}{m i n} \sum |f_{e} (p_{i})| + \sum |f_{s} (p_{i})| + \sum |f_{I} (p_{i})|

（20）

\{\begin{array}{l} p_{k} = [ρ_{k}, θ_{k}, z_{k}, I_{k}] \\ ρ_{k} = \sqrt[]{x_{k}^{2} + y_{k}^{2}} \\ θ_{k} = a r c t a n \frac{y_{k}}{x_{k}} \end{array}

（21）

S_{i j h} = \{p_{k} \in P |\frac{i L_{m a x}}{N_{r}} \leq ρ_{k} < \frac{(i + 1) L_{m a x}}{N_{r}}, \frac{j 2 π}{N_{s}} - π \leq θ_{k} < \frac{(j + 1) 2 π}{N_{s}} - π, \frac{h H_{m a x}}{N_{h}} \leq μ_{k} < \frac{(h + 1) H_{m a x}}{N_{h}}\}

（22）

\{\begin{array}{l} Z_{i j h} = \underset{p_{i} \in S_{i j h}}{m a x} z_{k} \\ I_{i j h} = \underset{p_{i} \in S_{i j h}}{m a x} I_{k} \end{array}

（23）

W_{i j h} = \{\begin{array}{l} 0, q = 0 \\ 1, Q_{i j h} > 2 q \\ \frac{Q_{i j h}}{2 q}, e l s e \end{array}

（24）

\{\begin{array}{l} D (i, j) = \sum_{h = 1}^{N_{h}} (Z_{i j h} + W_{i j h} I_{i j h}) \\ D = \cup_{i = 1, j = 1}^{N_{r}, N_{s}} D (i, j) \end{array}

（25）

d (D_{c}, D_{h i s t o r y}) = \frac{1}{N_{s}} \sum_{j = 1}^{N_{s}} (1 - \frac{c_{j, c} c_{j, h i s t o r y}}{‖c_{j, c}‖ ‖c_{j, h i s t o r y}‖})

（26）

S_{S I M} = \underset{i \in [1, N_{s}]}{m i n} d_{i} (D_{c}, D_{h i s t o r y})

（27）

Long Peng, Weigang Li, Qifeng Wang, Huan Yi. Lidar SLAM Algorithm Based on Point Cloud Geometry and Intensity Information[J]. Laser & Optoelectronics Progress, 2025, 62(8): 0815001

Download Citation

Set citation alerts for the article

Tools

Set citation alerts for the article

Save the article for my favorites

Paper Information

微信扫一扫：分享

微信扫一扫：分享