This is an implement of Fast Segmentation of 3D Point Clouds for Ground Vehicles [1].
The BEV module in toolbox is generously contributed by Ronny and Dr Kevin [2] [3].
We use a lidar image img\kitti_sample.pcd from the KITTI dataset as a test for this algorithm [4].
import pcl
import numpy as np
vel_msg = np.asarray(pcl.load('img/kitti_sample.pcd'))
vel_msg = vel_msg * np.array([1,1,-1]) # revert the z axisThe input of the module is numpy.ndarray type. Make sure to convert the data type when loading the cloud points.
Notes: the Velodyne LiDAR sensor is installed upside down in the NCLT and KITTI dataset.
from module.ground_removal import Processor
process = Processor(n_segments=70, n_bins=80, line_search_angle=0.3, max_dist_to_line=0.15,
sensor_height=1.73, max_start_height=0.5, long_threshold=8)
vel_non_ground = process(vel_msg)from module import lidar_projection
import cv2
img_raw = lidar_projection.birds_eye_point_cloud(vel_msg,
side_range=(-50, 50), fwd_range=(-50, 50),
res=0.25, min_height=-2, max_height=4)
cv2.imwrite('img/kitti_raw.png', img_raw)
img_non_ground = lidar_projection.birds_eye_point_cloud(vel_non_ground,
side_range=(-50, 50), fwd_range=(-50, 50),
res=0.25, min_height=-2, max_height=4)
cv2.imwrite('img/kitti_non_ground.png', img_non_ground)Module: Processor
Args:
n_segments(int): The number of fan-shaped regions divided by 360 degrees
n_bins(int): The number of bins divided in a segment.
r_max(float): The max boundary of lidar point.(meters)
r_min(float): The min boundary of lidar point.(meters)
line_search_angle(float): The angle for relative search in nearby segments.
max_dist_to_line(float): The distance threshold of the non-ground object to the ground.(meters)
max_slope(float): Local maximum slope of the ground.
max_error(float): The max MSE to fit the ground.(meters)
long_threshold(int): The max threshold of ground wave interval.
max_start_height(float): The max height difference between hillside and ground.(meters)
sensor_height(float): The distance from the lidar sensor to the ground.(meters)
Call:
Arg:
vel_msg(numpy.ndarray): The raw local LiDAR cloud points in 3D(x,y,z).
For example:
vel_msg shapes [n_point, 3], with `n_point` refers to the number of cloud points,
while `3` is the number of 3D(x,y,z) axis.
vel_msg = array([[0.3, 0.1, 0.7],
[0.6, 0.6, 0.5],
[0.1, 0.4, 0.8],
... ... ...
[0.5, 0.3, 0.6],
[0.6, 0.3, 0.4]]
Returns:
vel_non_ground(numpy.ndarray): The local LiDAR cloud points after filter out ground info.
Checkout to the Slope branch for the slope-free feature which removes both ground and the effect of the slope. The main difference between them is whether to enable the offset(offset)
An example can be found here
The main differences lie in,
- Is the lidar installed in reverse?
- The height of the lidar from the ground. (sensor_height)
[1] Himmelsbach, M., Hundelshausen, F.V. and Wuensche, H.J., 2010, June. Fast segmentation of 3D point clouds for ground vehicles. In 2010 IEEE Intelligent Vehicles Symposium (pp. 560-565). IEEE.
[2] Ronny., (2017). Lidar Birds Eye Views [online]. Ronny. Available from: http://ronny.rest/blog/post_2017_03_26_lidar_birds_eye
[3] Sun, L., Adolfsson, D., Magnusson, M., Andreasson, H., Posner, I. and Duckett, T., 2020, May. Localising Faster: Efficient and precise lidar-based robot localisation in large-scale environments. In 2020 IEEE International Conference on Robotics and Automation (ICRA) (pp. 4386-4392). IEEE.
[4] Geiger, A., Lenz, P., Stiller, C. and Urtasun, R., 2013. Vision meets robotics: The kitti dataset. The International Journal of Robotics Research, 32(11), pp.1231-1237.