TRaIL-Odom: Tightly Coupled Continuous Time Radar-IMU-LiDAR Odometry With Adaptive Doppler Weighting

TRaIL-Odom trajectory comparisons with reference paths and baseline methods in an airfield and bicycle tunnel.
Trajectory comparisons in open terrain and a tunnel, where directional geometric constraints are weak. Paper authors · Source

Abstract summary

TRaIL-Odom combines radar, inertial and LiDAR measurements in a continuous-time estimator while adapting the contribution of radar to the available geometric information. It first identifies translational directions that LiDAR constrains poorly. Individual Doppler measurements are then weighted according to how much information they provide along those directions. A second mechanism changes the overall radar influence as the geometry becomes more or less informative. This avoids assigning the same importance to radar in every scene and direction. Evaluation covers 13 sequences, including open areas and tunnels. On three degenerate sequences, the combined weighting mechanisms reduce absolute and relative trajectory errors by 86.0% and 78.5%, respectively, compared with the fixed-weight baseline.

Journal article · All publications