CAR-LOAM: color-assisted robust LiDAR odometry and mapping for solid-state LiDARs

LiDAR odometry and mapping (LOAM) has been playing an important role in robot perception due to its ability to simultaneously estimate the robot's pose and build high-precision maps of the surrounding environment. However, its accuracy inevitably degrades due to point correspondence outliers. This problem is more severe for solid-state LiDARs with irregular samplings. To tackle this problem, we propose CAR-LOAM, a visual-assisted LOAM framework designed for solid-state LiDARs that incorporates a perceptually uniform color difference weighting strategy to exclude color correspondence outliers and a robust error metric based on Welsch's function to suppress positional correspondence outliers. As a result, the system achieves accurate localization and reconstructs dense, precise, colored, and three-dimensional (3D) point cloud maps of the environment. Thorough experiments with challenging scenarios show that our method provides higher accuracy compared with current baselines and methods.