Granted Patent B1

US 12,493,726 · App. 19/067,753 · Granted Dec 9, 2025

Inventors: Lei Yang (Acworth, GA); Lisha Huang (Acworth, GA); Yongxian Zeng (Acworth, GA); Shifei Yan (Acworth, GA)

Assignee: VisionNav Robotics USA Inc.

G06F30/15G06F30/20

Abstract

Claims (74)

1 . A test method, comprising:

pre-acquiring handling device system parameters and simulation carrier model files, wherein each of the simulation carrier model files contains a simulation carrier model and carrier parameters, the simulation carrier models contained in different simulation carrier model files are different, and the carrier parameters contained in the different simulation carrier model files vary;

determining test cases according to the handling device system parameters and the simulation carrier model files;

generating simulation carrier point clouds corresponding to the determined test cases; and

in response to a simulation test request, calling a target test case in the determined test cases, and performing a simulation test on a to-be-tested perception algorithm according to a simulation carrier point cloud, of the simulation carrier point clouds, corresponding to the target test case to obtain a simulation test result of the to-be-tested perception algorithm, wherein

the performing of the simulation test on the to-be-tested perception algorithm according to the simulation carrier point cloud corresponding to the target test case to obtain the simulation test result of the to-be-tested perception algorithm comprises:

inputting the simulation carrier point cloud corresponding to the target test case into the to-be-tested perception algorithm;

identifying and obtaining a target relative pose of a simulation carrier model, of the simulation carrier models, relative to a simulation handling device model in the target test case from the simulation carrier point cloud corresponding to the target test case by the to-be-tested perception algorithm;

taking a reference relative pose of the simulation carrier model relative to the simulation handling device model in the target test case as a reference; and

determining the simulation test result of the to-be-tested perception algorithm according to a difference between the target relative pose and the reference relative pose.

2 . The method according to claim 1 ,

wherein the handling device system parameters comprise at least one of handling device parameters, sensor parameters, simulation scenario parameters, and a plurality of pose transformation strategies, and

wherein a simulation sensor described by the sensor parameters is configured on the simulation handling device model described by the handling device parameters, and wherein each of the plurality of pose transformation strategies is configured to indicate to transform a pose of the simulation handling device model or a pose of the simulation carrier model.

3 . The method according to claim 1 , wherein the determining of the test cases according to the handling device system parameters and the simulation carrier model files comprises:

selecting at least one simulation carrier model file from the simulation carrier model files;

selecting at least one set of handling device system parameters from the handling device system parameters, wherein the set of selected handling device system parameters contains at least one set of handling device parameters and at least one set of sensor parameters;

generating a test task according to the selected at least one simulation carrier model file and the selected handling device system parameters; and

performing permutation and combination on the at least one simulation carrier model file contained in the test task to obtain the test cases corresponding to the test task.

4 . The method according to claim 1 , wherein the determining of the test cases according to the handling device system parameters and the simulation carrier model files comprises:

in response to a selection operation of a user, determining at least one selected simulation carrier model file from the simulation carrier model files;

determining at least one set of selected handling device system parameters from the handling device system parameters, wherein the set of selected handling device system parameters contains at least one set of handling device parameters and at least one set of sensor parameters; and

generating the test cases according to the selected simulation carrier model files and the selected handling device system parameter.

5 . The method according to claim 3 , wherein the at least one set of selected handling device system parameters further contains at least one set of simulation scenario parameters and/or at least one set of pose transformation strategies.

6 . The method according to claim 1 , wherein

the handling device system parameters in the target test case comprises a handling device parameter and a sensor parameter, and

the generating of the simulation carrier point clouds corresponding to the determined test cases comprises:

for each of the determined test cases, determining an initial pose of the simulation carrier model according to the simulation carrier model and the carrier parameters contained in the selected at least one simulation carrier model file in the target test case;

determining an initial pose of the simulation handling device model according to the handling device parameters in the target test case;

determining the reference relative pose of the simulation carrier model relative to the simulation handling device model according to the initial pose of the simulation handling device model and the initial pose of the simulation carrier model; and

generating the simulation carrier point cloud corresponding to the target test case according to the sensor parameter in the target test case and the reference relative pose.

7 . The method according to claim 6 , wherein the handling device system parameter in the target test case contains comprises a pose transformation strategy, and

the determining of the reference relative pose of the simulation carrier model relative to the simulation handling device model according to the initial pose of the simulation handling device model and the initial pose of the simulation carrier model comprises:

when the pose transformation strategy in the target test case is a first pose transformation strategy, transforming the initial pose of the simulation handling device model, and determining the reference relative pose of the simulation handling device model relative to the simulation handling device model according to a transformed pose of the simulation handling device model and the initial pose of the simulation carrier model; and

when the pose transformation strategy in the target test case is a second pose transformation strategy, transforming the initial pose of the simulation carrier model, and determining the reference relative pose of the simulation carrier model relative to the simulation handling device model according to the initial pose of the simulation handling device model and a transformed pose of the simulation carrier model.

8 . The method according to claim 6 , wherein

the handling device system parameter in the target test case further contains a simulation scenario parameter, and

the generating of the simulation carrier point cloud corresponding to the target test case according to the sensor parameter in the target test case and the reference relative pose comprises:

generating the simulation carrier point cloud corresponding to the target test case according to the sensor parameter and the simulation scenario parameter in the target test case as well as the reference relative pose.

9 . The method according to claim 1 , wherein the target test case corresponds to a plurality of sets of the simulation carrier point clouds, and wherein the reference relative pose of the simulation carrier models corresponding to the plurality of sets of the simulation carrier point clouds relative to the simulation handling device model vary.

10 . The method according to claim 1 , wherein the performing of the simulation test on the to-be-tested perception algorithm according to the simulation carrier point cloud corresponding to the target test case to obtain the simulation test result of the to-be-tested perception algorithm comprises:

for each set of a plurality of sets of the simulation carrier point clouds corresponding to the target test case, inputting the set of the plurality of sets of the simulation carrier point clouds into the to-be-tested perception algorithm, and identifying and obtaining the target relative pose of the simulation carrier model corresponding to the set of the plurality of sets of the simulation carrier point clouds relative to the simulation handling device model from the set of the plurality of sets of the simulation carrier point clouds by the to-be-tested perception algorithm;

taking the reference relative pose of the simulation carrier model corresponding to the set of simulation carrier point clouds relative to the simulation handling device model as the reference, and determining the simulation test result of the to-be-tested perception algorithm for detecting the set of the plurality of sets of the simulation carrier point clouds according to the difference between the reference relative pose corresponding to the set of the plurality of sets of the simulation carrier point clouds and the target relative pose corresponding to the set of the plurality of sets of the simulation carrier point clouds; and

detecting the simulation test result of each set of the plurality of sets of the simulation carrier point clouds corresponding to the target test case according to the to-be-tested perception algorithm, and determining the simulation test result of the to-be-tested perception algorithm.

11 . The method according to claim 1 , wherein each set of the plurality of sets of the simulation carrier point clouds corresponds to a plurality of test distances, wherein a test distance of the plurality of test distances is defined as a distance between a carrier fork entry surface of the simulation carrier model and a center of a vehicle body of the simulation handling device model.

12 . The method according to claim 1 , wherein the to-be-tested perception algorithm comprises a plurality of perception algorithms, and each perception algorithm of the plurality of perception algorithms detects different features of a carrier when identifying a carrier pose.

13 . The method according to claim 1 , wherein the performing of the simulation test on the to-be-tested perception algorithm according to the simulation carrier point cloud corresponding to the target test case to obtain the simulation test result of the to-be-tested perception algorithm comprises:

for each to-be-tested perception algorithm, inputting the simulation carrier point cloud corresponding to the target test case into the to-be-tested perception algorithm, and identifying and obtaining the relative pose of the simulation carrier model in the target test case relative to the simulation handling device model from the simulation carrier point cloud corresponding to the target test case through detecting a feature of a carrier by the to-be-tested perception algorithm, wherein the relative pose is used as the target relative pose output by the to-be-tested perception algorithm; and

taking the reference relative pose of the simulation carrier model relative to the simulation handling device model in the target test case as the reference, and determining the simulation test result of the to-be-tested perception algorithm according to the difference between the target relative pose output by the to-be-tested perception algorithm and the reference relative pose.

14 . The method according to claim 1 , further comprising:

pre-acquiring a real carrier point cloud corresponding to the target test case;

in response to a real test request, testing the to-be-tested perception algorithm according to the real carrier point cloud corresponding to the target test case to obtain a real test result of the to-be-tested perception algorithm; and

determining a comprehensive test result of the to-be-tested perception algorithm according to the simulation test result and the real test result.

15 . The method according to claim 1 , wherein the pre-acquiring of the simulation carrier model files comprises:

pre-receiving the simulation carrier models input by a user and sizes corresponding to the simulation carrier models;

determining the carrier parameters according to the sizes corresponding to the simulation carrier models; and

selecting the simulation carrier model files according to the simulation carrier models and the carrier parameters.

16 . The method according to claim 1 , further comprising:

statistically analyzing simulation test results of the to-be-tested perception algorithm, and determining a precision of the to-be-tested perception algorithm for detecting a pose of the simulation carrier model corresponding to the target test case; and

generating a test report corresponding to the to-be-tested perception algorithm according to the determined precision of the to-be-tested perception algorithm for detecting the pose of the simulation carrier model corresponding to the target test case.

17 . The method according to claim 1 , wherein the performing of the simulation test on the to-be-tested perception algorithm according to the simulation carrier point cloud corresponding to the target test case to obtain the simulation test result of the to-be-tested perception algorithm comprises:

acquiring a preset query period and a preset test timeout duration;

when the query period is reached, querying whether there is the simulation test result of the to-be-tested perception algorithm obtained by executing the target test case, if yes, updating a test state of the target test case to be completed, and displaying the updated test state of the target test case to a user, if no, continuing to wait; and

when the preset test timeout duration is reached, and a query finds that there is no simulation test result of the to-be-tested perception algorithm obtained by executing the target test case, updating the test state of the target test case to be abnormal, and displaying the abnormal test state of the target test case to the user.

18 . A non-transitory computer-readable storage medium, storing a computer program which, when executed by a processor, causes the processor to perform operations, comprising:

pre-acquiring handling device system parameters and simulation carrier model files, wherein each of the simulation carrier model files contains a simulation carrier model and carrier parameters, the simulation carrier models contained in different simulation carrier model files are different, and the carrier parameters contained in the different simulation carrier model files vary;

determining test cases according to the handling device system parameters and the simulation carrier model files;

generating simulation carrier point clouds corresponding to the determined test cases; and

in response to a simulation test request, calling a target test case in the determined test cases, and performing a simulation test on a to-be-tested perception algorithm according to a simulation carrier point cloud, of the simulation carrier point clouds, corresponding to the target test case to obtain a simulation test result of the to-be-tested perception algorithm, wherein

the performing of the simulation test on the to-be-tested perception algorithm according to the simulation carrier point cloud corresponding to the target test case to obtain the simulation test result of the to-be-tested perception algorithm comprises:

inputting the simulation carrier point cloud corresponding to the target test case into a to-be-tested perception algorithm;

identifying and obtaining a target relative pose of a simulation carrier model relative to a simulation handling device model in the target test case from the simulation carrier point cloud corresponding to the target test case by the to-be-tested perception algorithm;

taking a reference relative pose of the simulation carrier model relative to the simulation handling device model in the target test case as a reference; and

determining the simulation test result of the to-be-tested perception algorithm according to a difference between the target relative pose and the reference relative pose.

19 . The method of claim 2 , wherein the simulation handling device model described by the handling device parameters includes at least one from a group of an automated guided vehicle (AGV), an autonomous mobile robot (AMR), a humanoid robot, a forklift, a reach truck, a stacker, and a tow tractor.

Assignments (1)

Priority Claims (1)

References Cited (11)

Cited By (1)

US 12,606,427