FAST-LIVO2 on Resource-Constrained Platforms: LiDAR-Inertial-Visual Odometry with Efficient Memory and Computation

TL;DR

This work tackles the challenge of running LiDAR-inertial-visual odometry (LIVO) on resource-constrained platforms by introducing a degeneration-aware adaptive visual frame selector integrated into the error-state iterated Kalman filter (ESIKF) and a memory-efficient hybrid map that couples a compact local map with a long-term visual map. The LiDAR degeneration evaluation uses normalized singular values from a plane-based residual model to detect degeneracy, which drives adaptive image selection and reduces computational load while preserving robustness. A unified map structure with a hash-based root voxel and an octree-based local map, complemented by a long-term visual map, yields substantial memory savings without sacrificing accuracy. Extensive experiments on Hilti public datasets and challenging private sequences demonstrate significant improvements in computation efficiency (up to ~33% per-frame runtime reduction) and memory usage (up to ~47% reduction) with only modest declines in RMSE, and real-time performance on a low-cost ARM platform, validating scalable edge deployments for LIVO.

Abstract

This paper presents a lightweight LiDAR-inertial-visual odometry system optimized for resource-constrained platforms. It integrates a degeneration-aware adaptive visual frame selector into error-state iterated Kalman filter (ESIKF) with sequential updates, improving computation efficiency significantly while maintaining a similar level of robustness. Additionally, a memory-efficient mapping structure combining a locally unified visual-LiDAR map and a long-term visual map achieves a good trade-off between performance and memory usage. Extensive experiments on x86 and ARM platforms demonstrate the system's robustness and efficiency. On the Hilti dataset, our system achieves a 33% reduction in per-frame runtime and 47% lower memory usage compared to FAST-LIVO2, with only a 3 cm increase in RMSE. Despite this slight accuracy trade-off, our system remains competitive, outperforming state-of-the-art (SOTA) LIO methods such as FAST-LIO2 and most existing LIVO systems. These results validate the system's capability for scalable deployment on resource-constrained edge computing platforms.

Figures (11)

• Figure 1: Overview of a real-time experiment on the low-power ARM platform. (a) The ARM platform RK3588 with an octa-core architecture (4× Cortex-A76 + 4× Cortex-A55) and a maximum frequency of 2.4GHz used for testing, priced at approximately 100 USD. (b) Point cloud of the nighttime street scene used for testing, with the orange trajectory representing the collected path. (c) Detailed per-frame runtime statistics of the system, with data input from both LiDAR and camera at 10 Hz (100 ms per frame).
• Figure 2: System overview. In the figure detailing voxel data on the lower-right side, the dashed ellipse encloses points considered to lie on a plane with varying scales.
• Figure 3: Illustration of sequential update ESIKF with adaptive visual frame selector.
• Figure 4: Illustration of map structure and sliding process in our system and FAST-LIVO2.
• Figure 5: Our platform with hardware synchronization for data acquisition.