Search papers, labs, and topics across Lattice.
GenZ-LIO is proposed to improve LiDAR-inertial odometry (LIO) robustness and efficiency across indoor and outdoor environments by addressing variations in spatial scale and point density. The framework adaptively regulates voxel size using a PID-inspired controller, employs a hybrid-metric state update combining point-to-plane and point-to-point residuals, and introduces a voxel-pruned correspondence search to reduce computational load. Experiments demonstrate that GenZ-LIO achieves robust odometry estimation and improved computational efficiency in diverse environments.
GenZ-LIO tackles the LiDAR-inertial odometry challenge of spatial scale variation, enabling robust and efficient localization across indoor and outdoor environments.
Light detection and ranging (LiDAR)-inertial odometry (LIO) enables accurate localization and mapping for autonomous navigation in various scenes. However, its performance remains sensitive to variations in spatial scale, which refers to the spatial extent of the scene reflected in the distribution of point ranges in a LiDAR scan. Transitions between confined indoor and expansive outdoor spaces induce substantial variations in point density, which may reduce robustness and computational efficiency. To address this issue, we propose GenZ-LIO, a LIO framework generalizable across both indoor and outdoor environments. GenZ-LIO comprises three key components. First, inspired by the principle of the proportional-integral-derivative (PID) controller, it adaptively regulates the voxel size for downsampling via feedback control, driving the voxelized point count toward a scale-informed setpoint while enabling stable and efficient processing across varying scene scales. Second, we formulate a hybrid-metric state update that jointly leverages point-to-plane and point-to-point residuals to mitigate LiDAR degeneracy arising from directionally insufficient geometric constraints. Third, to alleviate the computational burden introduced by point-to-point matching, we introduce a voxel-pruned correspondence search strategy that discards non-promising voxel candidates and reduces unnecessary computations. Experimental results demonstrate that GenZ-LIO achieves robust odometry estimation and improved computational efficiency across confined indoor, open outdoor, and transitional environments. Our code will be made publicly available upon publication.
