BREAKING
LIMOT goes open source
Loosely vs tightly coupled
Conventionalloose
●SLAM and MOT split
●Pose errors hurt tracking
LIMOTtight
●Joint optimization
●Ego and object poses together
How LIMOT works
1LiDAR de-skew
↓
2IMU pre-integration
↓
3Data association
↓
4Factor graph opt
Built on DL-SLOT plus IMU
Setup and license
Runsetup
●Ubuntu 20.04 + ROS Noetic
●GTSAM 4.0.3, KITTI data
Termslicense
●BSD-3-Clause
●Needs external 3D detector
On GitHub and IEEE RAL
AI NEWS BLITZ
A tightly-coupled LiDAR-inertial odometry and tracking system is now open source.