Event
Decentralized LiDAR-Radar-Inertial Odometry Framework Proposed
1 reports · 1 sources
Moumita Mukherjee and other authors proposed a decentralized LiDAR-radar-inertial odometry framework, LRIO, which fuses LiDAR, FMCW radar, and inertial sensing to estimate poses in GPS-denied underground tunnels affected by smoke. The framework uses an adaptive residual threshold (ART) to detect sensor degradation over time and adjust fusion, together with a fault-tolerant optimal information filter (RETOIF) to fuse LiDAR-inertial and multi-channel radar-inertial odometry in a fault-aware manner. The authors evaluated it using real sensor data from underground tunnels under two conditions: smoke-free and smoke-degraded. In the smoke-free experiment, RETOIF's positioning root mean square error was 2.1733, versus 2.9929 for conventional centralized LRIO, which the authors report as a 27.4% reduction.
Report timeline
- LRIO: Resilient Decentralized Radar-LiDAR-Inertial Odometry for Subterranean Environments
The authors propose LRIO, a decentralized LiDAR-radar-inertial odometry framework that fuses LiDAR, FMCW radar, and inertial sensing for pose estimation in GPS-denied subterranean tunnels affected by smoke. At the core of the framework is an Adaptive Residual Thresholding (ART) mechanism, which detects sensor degradation over time and adjusts the fusion process, combined with a Robust Error-Tolerant Optimal Information Filter (RETOIF) that fuses LiDAR-inertial and multiple radar-inertial odometry estimates in a fault-aware manner.
arXiv cs.RO ·