LaneSLAM

Semantic SLAM designed for urban road scene.

Tightly Coupled Lidar-IMU Inertial Mapping Framework

1.calibrate lidar and imu, lidar and camera

2.color the point cloud.

Semantics Based on LIDAR intensity and Image segmentation

1.Road segmentation based on deep learning.

2.Extract Ground cloud and Road Points(Line, road mark , crosswalk、curb and other road elements)

• Ground extraction based on Ransanc and plane fitting
• Use intensity-based filter to extract road points, maybe you need to calibrate the intensity of your lidar first and rescale your intensity to a reasonable scale.

3.Fuse lidar points with images and get a semantic map.

Object SLAM based on Geometric and metric information

1.Multi-frame Line tracking and fitting.

2.Stop Line based loop clouser