Degradation Resilient LiDAR-Radar-Inertial Odometry

Morten Nissov,Nikhil Khedekar,Kostas Alexis

2024-03-08

Abstract:Enabling autonomous robots to operate robustly in challenging environments is necessary in a future with increased autonomy. For many autonomous systems, estimation and odometry remains a single point of failure, from which it can often be difficult, if not impossible, to recover. As such robust odometry solutions are of key importance. In this work a method for tightly-coupled LiDAR-Radar-Inertial fusion for odometry is proposed, enabling the mitigation of the effects of LiDAR degeneracy by leveraging a complementary perception modality while preserving the accuracy of LiDAR in well-conditioned environments. The proposed approach combines modalities in a factor graph-based windowed smoother with sensor information-specific factor formulations which enable, in the case of degeneracy, partial information to be conveyed to the graph along the non-degenerate axes. The proposed method is evaluated in real-world tests on a flying robot experiencing degraded conditions including geometric self-similarity as well as obscurant occlusion. For the benefit of the community we release the datasets presented:

Robotics,Signal Processing

What problem does this paper attempt to address?

The paper aims to address the problem of robust navigation for autonomous robots in challenging environments. Specifically, the study proposes a tightly coupled fusion method of LiDAR, radar, and Inertial Measurement Unit (IMU) to improve the accuracy and robustness of pose estimation in GPS-denied environments, such as those with geometric self-similarity or obstructions like smoke and dust.

The core contributions of the paper include:
1. Proposing a LiDAR-Radar-Inertial Odometry method (DR-LRIO) for self-similar environments and degraded conditions, utilizing radar data to compensate for the limitations of LiDAR under these conditions.
2. Introducing new factor formulations within the factor graph framework to directly integrate LiDAR features into the global map and fuse the linear velocity information estimated by radar into the factor graph.
3. Validating the effectiveness of the proposed method through a series of experiments, including tests in a motion capture field, geometrically self-similar tunnels, and fog-filled corridors, demonstrating performance improvements under various degraded conditions.

In summary, the study aims to develop a multimodal sensor fusion scheme capable of handling perception degradation to enhance the localization and navigation capabilities of autonomous robots in complex environments.