Method and apparatus for constructing reflectance map

Patent metadata

Field	Value
Publication number	US-10627520-B2
Application number	US-201715800441-A
Country	US
Kind code	B2
Filing date	Nov 1, 2017
Priority date	Aug 15, 2017
Publication date	Apr 21, 2020
Grant date	Apr 21, 2020

Abstract

A specific implementation of the method includes: constructing a reflectance map based on a position and an Euler angle, obtained through a global pose optimization and used for constructing a reflectance map, of a center of a laser radar corresponding to each frame laser point cloud used for constructing the reflectance map. This implementation implements the level-by-level pose optimization of key frame laser point clouds, sample frame laser point clouds, regular frame laser point clouds selected from laser point clouds used for constructing a reflectance map, to obtain an accurate position and Euler angle, used for constructing the reflectance map, of a center of the laser radar corresponding to each frame laser point cloud used for constructing the reflectance map, so that accurate coordinates of laser points in each frame laser point cloud used for constructing the reflectance map in a world coordinate system can be obtained.

First claim

What is claimed is:

1. A method for constructing a reflectance map, comprising:
selecting, from laser point clouds collected in a region corresponding to a to-be-constructed reflectance map, laser point clouds used for constructing a reflectance map, and selecting sample frame laser point clouds from the laser point clouds used for constructing the reflectance map;
selecting key frame laser point clouds from the sample frame laser point clouds, and determining an optimal key frame laser point cloud based on adjustment amounts of the key frame laser point clouds, each of the adjustment amounts being determined based on an amount of movement between a center position of a laser radar corresponding to a key frame laser point cloud after being merged with a second key frame laser point cloud and a center position of the laser radar corresponding to the key frame laser point cloud;
performing a global pose optimization on laser point clouds other than the optimal key frame laser point cloud in the laser point clouds used for constructing the reflectance map, to obtain a position and an Euler angle, used for constructing the reflectance map, of a center of the laser radar corresponding to each frame laser point clouds used for constructing the reflectance map; and
constructing the reflectance map based on the position and the Euler angle, used for constructing the reflectance map, of the center of the laser radar corresponding to each frame laser point clouds used for constructing the reflectance map.

2. The method according to claim 1 , wherein the selecting, from laser point clouds collected in a region corresponding to a to-be-constructed reflectance map, laser point clouds used for constructing a reflectance map comprises:
removing laser point clouds having a collection time with an erroneous timestamp, from the laser point clouds collected in the region corresponding to the to-be-constructed reflectance map;
removing laser point clouds having identical center positions of the laser radar, from the laser point clouds collected in the region corresponding to the to-be-constructed reflectance map; and
using the remaining laser point clouds in the laser point clouds collected in the region corresponding to the to-be-constructed reflectance map, as the laser point clouds used for constructing a reflectance map.

3. The method according to claim 2 , wherein the determining an optimal key frame laser point cloud based on adjustment amounts of the key frame laser point clouds comprises:
calculating an average adjustment amount of each of the key frame laser point clouds, wherein the average adjustment amount is obtained by dividing a sum of the adjustment amounts corresponding to the key frame laser point cloud by a number of other key frame laser point clouds merged with the key frame laser point clouds;
determining a key frame laser point cloud having a greatest average adjustment amount, and recalculating average adjustment amounts of remaining key frame laser point clouds other than the key frame laser point clouds having the greatest average adjustment amount; and
assigning a key frame laser point clouds having a smallest average adjustment amount from the remaining key frame laser point clouds as the optimal key frame laser point cloud.

4. The method according to claim 3 , wherein the performing a global pose optimization on laser point clouds other than the optimal key frame laser point clouds in the laser point clouds used for constructing the reflectance map comprises:
using a position and an Euler angle of a center of the laser radar corresponding to the optimal key frame laser point cloud as the position and an Euler angle, used for constructing the reflectance map, of the center of the laser radar corresponding to the optimal key frame laser point cloud, and for each of other key frame laser point clouds other than the optimal key frame laser point cloud, performing the pose optimization to obtain a position and an Euler angle, used for constructing the reflectance map, of a center of the laser radar corresponding to each of the other key frame laser point clouds;
for each of other sample frame laser point clouds other than the key frame laser point cloud in the sample frame laser point clouds, performing the pose optimization to obtain a position and an Euler angle, used for constructing the reflectance map, of a center of the laser radar corresponding to each of the other sample frame laser point clouds; and
for each of regular frame laser point clouds other than the sample frame laser point clouds in the laser point clouds used for constructing the reflectance map, performing the pose optimization to obtain a position and an Euler angle, used for constructing the reflectance map, of a center of the laser radar corresponding to each of the regular frame laser point clouds.

5. The method according to claim 4 , wherein the for each of other key frame laser point clouds other than the optimal key frame of laser point cloud, performing the pose optimization comprises:
calculating, based on a constraint condition corresponding to the other key frame laser point clouds, an optimized position and an optimized Euler angle of a center of the laser radar corresponding to each of the other key frame laser point clouds that satisfy a convergence condition corresponding to the other key frame laser point clouds, wherein the constraint condition corresponding to the other key frame laser point clouds comprises: the position and the Euler angle of the center of the laser radar corresponding to each of the other key frame laser point clouds, a weight corresponding to the position and the Euler angle of the center of the laser radar corresponding to each of the other key frame laser point clouds, and a transformation relationship between the other key frame laser point clouds;
respectively calculating an optimization result and a merging result of each pair of the other key frame laser point clouds, wherein a pair of the other key frame laser point clouds comprises two different other key frame laser point clouds, the optimization result of the pair of the other key frame laser point clouds is an average value of differentials between optimized center positions of the laser radar corresponding to the other key frame laser point clouds in the pair of the other key frame laser point clouds and center positions of the laser radar corresponding to the other key frame laser point clouds, and the merging result of the pair of the other key frame laser point clouds is an amount of movement between a center position of the laser radar obtained after each of the pair of the other key frame laser point clouds is merged with the other one of the pair of the other key frame laser point clouds, and the center position of the laser radar corresponding to the each of the pair of the other key frame laser point clouds;
removing, from the constraint condition corresponding to the other key frame laser point clouds, a transformation relationship between two other key frame laser point clouds in the pair of the other key frame laser point clouds having the corresponding optimization result and merging result with differentials greater than a threshold, to obtain a new constraint condition;
calculating, according to the new constraint condition, the optimized position and Euler angle of the center of the laser radar corresponding to each of the other key frame laser point clouds that satisfy the convergence condition corresponding to the other key frame laser point clouds; and
using the optimized position and Euler angle of the center of the laser radar corresponding to each of the other key frame laser point clouds as the position and the Euler angle, used for constructing the reflectance map, of the center of the laser radar corresponding to th

Assignees

• Baidu online network technology beijing co ltd

Inventors

• Yu Li
• Yi Shichun
• Song Shiyu
• Dong Fangfang
• Xu Baoqiang

Classifications

• G06T2207/10028

  Range image; Depth image; 3D point clouds · CPC title

• G01S17/89 Primary

  for mapping or imaging · CPC title

• G01S7/4808

  Evaluating distance, position or velocity data · CPC title

• G01S7/4802 Primary

  using analysis of echo signal for target characterisation; Target signature; Target cross-section · CPC title

• G06T7/70

  Determining position or orientation of objects or cameras (camera calibration G06T7/80) · CPC title