Method, apparatus, computing device and computer-readable storage medium for positioning

Patent metadata

Field	Value
Publication number	US-11725944-B2
Application number	US-202016806331-A
Country	US
Kind code	B2
Filing date	Mar 2, 2020
Priority date	Mar 2, 2020
Publication date	Aug 15, 2023
Grant date	Aug 15, 2023

Abstract

The present disclosure provides a method, an apparatus, a computer device and a computer-readable storage medium for positioning, and relates to the field of autonomous driving. The method obtains inertial measurement data of a device to be positioned at a current time and point cloud data collected by a LiDAR on the device at the current time; determines, by integrating the inertial measurement data, inertial positioning information of the device in an inertial coordinate system at the current time; and determines, based on the inertial positioning information, the point cloud data and at least one local map built in a local coordinate system, a positioning result of the device in the local coordinate system at the current time. Techniques of the present disclosure can provide an effective and stable local positioning result.

First claim

We claim:

1. A method, comprising:
obtaining inertial measurement data of a device at a first time and point cloud data collected by a LiDAR on the device at the first time;
determining, by integrating the inertial measurement data, inertial positioning information of the device in an inertial coordinate system at the first time based on the inertial measurement data; and
determining, based on the inertial positioning information, the point cloud data and at least one local map built in a local coordinate system, a first positioning result of the device in the local coordinate system at the first time,
wherein the at least one local map is built based on a plurality of frames of historical point cloud data collected by the LiDAR on the device, and
wherein the determining the first positioning result comprises:
determining a first posterior probability associated with the first positioning result based on a second positioning result of the device at a second time prior to the first time, the point cloud data, the inertial positioning information, and the at least one local map; and
determining the first positioning result by maximizing the first posterior probability.

2. The method of claim 1 , wherein the determining the first posterior probability comprises:
determining a first likelihood value of the point cloud data with respect to the first positioning result and the at least one local map;
determining a second likelihood value of the inertial positioning information with respect to the first positioning result and the second positioning result; and
determining, based on the first likelihood value and the second likelihood value, the first posterior probability.

3. The method of claim 2 , wherein the at least one local map comprises a plurality of local maps having different resolutions, and the determining the first likelihood value comprises:
determining, for a local map of the plurality of local maps, a likelihood value of the point cloud data with respect to the first positioning result and the local map; and
determining the first likelihood value based on a plurality of likelihood values determined for the plurality of local maps.

4. The method of claim 2 , wherein:
the point cloud data comprises respective reflection information of a plurality of laser points,
the at least one local map comprises a 3D local map, the 3D local map including a plurality of grids, each grid having corresponding laser reflection information and obstacle occupancy probability, and
the determining the first likelihood value comprises:
determining, from the plurality of grids, a group of grids hit by the plurality of laser points by matching the point cloud data with the 3D local map; and
determining, based on a group of obstacle occupancy probabilities corresponding to the group of grids, laser reflection information corresponding to the group of grids and respective reflection information of the plurality of laser points in the point cloud data, the first likelihood value of the point cloud data with respect to the first positioning result and the 3D local map.

5. The method of claim 1 , further comprising:
prior to the determining the first positioning result, performing motion compensation on the point cloud data based on the inertial positioning information.

6. The method of claim 1 , further comprising:
in response to the first positioning result being determined, optimizing the first positioning result based on at least the inertial positioning information.

7. The method of claim 6 , wherein the first positioning result includes a relative pose of the point cloud data relative to the at least one local map, a first pose of the device in the local coordinate system and a second pose of the at least one local map in the local coordinate system, and the optimizing the first positioning result comprises:
optimizing the first pose and the second pose while keeping the relative pose unchanged.

8. The method of claim 7 , wherein the optimizing the first positioning result comprises:
determining a second posterior probability associated with a group of positioning results of the device, wherein the group of positioning results comprises at least the first positioning result of the device at the first time and a second positioning result of the device in the local coordinate system at a second time prior to the first time; and
optimizing the first positioning result by maximizing the second posterior probability.

9. The method of claim 8 , wherein the determining the second posterior probability comprises:
determining a third likelihood value associated with the first positioning result;
determining a fourth likelihood value of the inertial positioning information with respect to the first positioning result and the second positioning result; and
determining the second posterior probability based on at least the third likelihood value and the fourth likelihood value.

10. The method of claim 9 , wherein the determining the third likelihood value comprises:
determining, based on the first pose and the second pose, an estimate for the relative pose;
determining a residual between the estimate and the relative pose indicated by the first positioning result; and
determining, based on at least the residual, the third likelihood value of the relative pose with respect to the first pose and the second pose.

11. The method of claim 9 , wherein the determining the second posterior probability comprises:
determining a fifth likelihood value associated with the second positioning result;
determining a sixth likelihood value associated with a second inertial positioning information of the device in the inertial coordinate system at the second time; and
determining the second posterior probability based on at least the third likelihood value, the fourth likelihood value, the fifth likelihood value and the sixth likelihood value.

12. The method of claim 1 , further comprising:
updating the at least one local map based on the point cloud data.

13. A computing device, comprising:
one or more processors; and
a memory for storing one or more programs, which, when executed by the one or more processors, cause the computing device to perform acts including:
obtaining inertial measurement data of a device at a first time and point cloud data collected by a LiDAR on the device at the first time;
determining, by integrating the inertial measurement data, inertial positioning information of the device in an inertial coordinate system at the first time; and
determining, based on the inertial positioning information, the point cloud data and at least one local map built in a local coordinate system, a first positioning result of a first pose of the device in the local coordinate system at the first time,

wherein the at least one local map is built based on a plurality of frames of historical point cloud data collected by the LiDAR on the device, and
wherein the determining the first positioning result comprises:
determining a first posterior probability associated with the first positioning result based on a second positioning result of the device at a second time prior to the first time, the point cloud data, the inertial positioning information, and the at least one local map; and
determining the first positioning result maximizing the first posterior probability.

14. The computing device of claim 13 , wherein the determining the first posterior probability comprises:
determining a first likelihood value of the point cloud data with respect to the first positioning result and the at least one local map;
deter

Assignees

• Apollo Intelligent Driving Technology Beijing Co Ltd

Inventors

• Hou Shenhua
• Ding Wendong
• Gao Hang
• Wan Guowei
• Song Shiyu

Classifications

• G01S7/4808

  Evaluating distance, position or velocity data · CPC title

• G01S17/86

  Combinations of lidar systems with systems other than lidar, radar or sonar, e.g. with direction finders · CPC title

• G01S17/06

  Systems determining position data of a target · CPC title

• G01C21/3602

  Input other than that of destination using image analysis, e.g. detection of road signs, lanes, buildings, real preceding vehicles using a camera · CPC title

• G01C21/1652 Primary

  with ranging devices, e.g. LIDAR or RADAR · CPC title