CN112052300A - SLAM back-end processing method, device and computer readable storage medium - Google Patents
SLAM back-end processing method, device and computer readable storage medium Download PDFInfo
- Publication number
- CN112052300A CN112052300A CN202010778658.0A CN202010778658A CN112052300A CN 112052300 A CN112052300 A CN 112052300A CN 202010778658 A CN202010778658 A CN 202010778658A CN 112052300 A CN112052300 A CN 112052300A
- Authority
- CN
- China
- Prior art keywords
- pose
- sequence
- line segments
- point cloud
- cloud data
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000003672 processing method Methods 0.000 title claims abstract description 21
- 238000003860 storage Methods 0.000 title claims abstract description 12
- 238000000034 method Methods 0.000 claims abstract description 44
- 238000012545 processing Methods 0.000 claims abstract description 36
- 238000005457 optimization Methods 0.000 claims abstract description 32
- 230000004044 response Effects 0.000 claims abstract description 16
- 239000011159 matrix material Substances 0.000 claims description 25
- 230000009466 transformation Effects 0.000 claims description 20
- 230000008859 change Effects 0.000 claims description 16
- 238000004590 computer program Methods 0.000 claims description 12
- 230000001172 regenerating effect Effects 0.000 claims description 3
- 230000036544 posture Effects 0.000 claims 15
- 238000013507 mapping Methods 0.000 abstract description 9
- 230000008569 process Effects 0.000 description 8
- 238000010586 diagram Methods 0.000 description 6
- 238000001514 detection method Methods 0.000 description 4
- 238000010276 construction Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 238000001914 filtration Methods 0.000 description 3
- 230000008030 elimination Effects 0.000 description 2
- 238000003379 elimination reaction Methods 0.000 description 2
- 230000035772 mutation Effects 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 206010063385 Intellectualisation Diseases 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008094 contradictory effect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 238000011065 in-situ storage Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
- 230000001360 synchronised effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/29—Geographical information databases
Abstract
The application discloses a SLAM back-end processing method, a device and a computer readable storage medium, wherein the method comprises the following steps: receiving map data to be processed and response data, wherein the map data to be processed comprises point cloud data and a pose sequence corresponding to the point cloud data, and the response data comprises information of a plurality of line segments; adjusting the positions of the multiple line segments based on the information of the multiple line segments and the first point cloud data located near the multiple line segments to obtain second point cloud data located near the adjusted multiple line segments and corresponding first position and posture sequences; adjusting the first position sequence to obtain a second position sequence; processing the second posture sequence to obtain a third posture sequence so as to ensure that the third posture sequence is continuous; and performing combined optimization based on the third posture sequence and the received preset constraint to obtain new map data. By the mode, the SLAM mapping consistency can be improved.
Description
Technical Field
The present application relates to the field of image processing technologies, and in particular, to a method and an apparatus for performing SLAM backend processing, and a computer-readable storage medium.
Background
The mobile robot is a robot with high self-planning, self-organization and self-adaptation capabilities, and the mobile robot has shown increasingly wide application prospects in various industries at present; in the research of the related technologies of the mobile robot, the navigation positioning technology is the core and key of the research and is also the key for realizing the intellectualization of the mobile robot, and the map building is the important basis for realizing the positioning of the robot, so that the robot needs to obtain an accurate map firstly and then combines the sensor data to realize the real-time positioning to realize the functions of automatic driving and automatic navigation.
At present, two-dimensional laser-based synchronous positioning And Mapping (SLAM) schemes are mainly divided into a filtering-based method And a graph-based optimization method, And both the two methods cannot completely ensure the graph building effect, so that solutions related to manual processing are developed successively, And the solutions can be divided into two types correspondingly: the method for optimizing the constructed image based on the filtering and the method for optimizing the constructed image based on the image optimization have no closed loop detection, so that errors are gradually accumulated in the process of constructing the image, and when the range of constructing the image is large, the large errors are generated; the elimination of accumulated errors in the map construction optimization method based on map optimization is mainly realized by closed-loop detection of a front end part, however, the closed-loop detection cannot completely guarantee the correctness of each detection, the wrong closed loop can seriously influence the rear end to generate a correct result, and the positioning accuracy and the map consistency are obviously reduced.
Disclosure of Invention
The application provides a SLAM back-end processing method, a device and a computer readable storage medium, which can improve SLAM mapping consistency.
In order to solve the above technical problem, a technical solution adopted by the present application is to provide a method for processing a SLAM backend, including: receiving map data to be processed and response data, wherein the map data to be processed comprises point cloud data and a pose sequence corresponding to the point cloud data, and the response data comprises information of a plurality of line segments; adjusting the positions of the multiple line segments based on the information of the multiple line segments and the first point cloud data located near the multiple line segments to obtain second point cloud data located near the adjusted multiple line segments and corresponding first position and posture sequences; adjusting the first position sequence to obtain a second position sequence; processing the second posture sequence to obtain a third posture sequence so as to ensure that the third posture sequence is continuous; and performing combined optimization based on the third posture sequence and the received preset constraint to obtain a new map number.
In order to solve the above technical problem, another technical solution adopted by the present application is to provide a SLAM backend processing apparatus, which includes a memory and a processor connected to each other, wherein the memory is used for storing a computer program, and the computer program is used for implementing the SLAM backend processing method when being executed by the processor.
In order to solve the above technical problem, another technical solution adopted by the present application is to provide a computer-readable storage medium for storing a computer program, wherein the computer program is configured to implement the SLAM backend processing method when executed by a processor.
Through the scheme, the beneficial effects of the application are that: the map data to be processed can be obtained firstly, and when the map data to be processed is judged manually to determine that an error exists in the map, a corresponding response data is generated by manually inputting an instruction; then adjusting the positions of a plurality of line segments in the response data so that the plurality of line segments are closest to the point cloud data nearby the adjusted line segments; then adjusting the pose sequence corresponding to the adjusted line segments, and eliminating discontinuity in the pose sequence after pose adjustment; then, performing joint optimization according to the generated pose sequence and a preset constraint added manually to generate new map data; the existing processing scheme is generally prone to online processing, while the map is processed in an offline mode, the map is irrelevant to a mapping method, the pose map can be adjusted through manually selected line segments and added constraints in the later period, and then back-end optimization is carried out, so that the wrong part in the map is repaired, and the consistency of laser SLAM mapping is improved.
Drawings
In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed to be used in the description of the embodiments are briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present application, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without creative efforts. Wherein:
fig. 1 is a schematic flowchart of an embodiment of a SLAM backend processing method provided in the present application;
fig. 2 is a schematic flowchart of another embodiment of a SLAM backend processing method provided by the present application;
fig. 3 is a schematic diagram of pose adjustment in the embodiment of fig. 2;
FIG. 4 is a schematic diagram of elimination of a sudden pose change in the embodiment of FIG. 2;
fig. 5 is a schematic structural diagram of an embodiment of a SLAM backend processing apparatus provided by the present application;
FIG. 6 is a schematic structural diagram of an embodiment of a computer-readable storage medium provided in the present application.
Detailed Description
The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only a part of the embodiments of the present application, and not all the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.
Referring to fig. 1, fig. 1 is a schematic flowchart illustrating an embodiment of a method for processing a SLAM backend provided by the present application, where the method includes:
step 11: and receiving the map data to be processed and the response data.
The map data to be processed sent by other devices can be received, or the map data to be processed is read from the storage device, or the map data to be processed input by a user can be received, wherein the map data to be processed comprises point cloud data and a pose sequence corresponding to the point cloud data; after the map data to be processed is acquired, whether an error part exists can be judged manually, and if the map data to be processed does not have the error part, a map can be directly generated; if the map data to be processed has an error part, the error part needs to be corrected, and the accuracy of the generated map is ensured.
Further, corresponding response data can be generated based on input of a user, the response data comprises information of a plurality of line segments, the plurality of line segments can be obtained by manually judging the acquired map data to be processed, and the selected line segments which possibly have errors and need to be adjusted are selected; for example, the map data to be processed may be displayed on a display device, and a user may manually draw a plurality of line segments in the point cloud data; for example, a user may draw three line segments that are approximately in the same direction, which may be merged into one line segment.
Step 12: and adjusting the positions of the line segments based on the information of the line segments and the first point cloud data near the line segments to obtain second point cloud data near the adjusted line segments and corresponding first position and posture sequences.
Because when a plurality of line segments are manually drawn in the point cloud data, the line segments may be different from the point cloud data which is actually required to be adjusted, after the response data is generated, the positions of the line segments can be adjusted according to the information of the manually selected line segments and the first point cloud data around the line segments, so that the second point cloud data around the adjusted line segments and the adjusted line segments are best fit.
Step 13: and adjusting the first position and posture sequence to obtain a second position and posture sequence.
After the positions of the line segments are adjusted, the pose can be further adjusted according to the relative pose transformation relation among the first pose sequences corresponding to the adjusted line segments; specifically, the number of line segments drawn by the user is two, the first pose sequence corresponding to one line segment can be set to be fixed, the first pose sequence corresponding to the other line segment is adjusted according to the pose transformation relation and the first pose sequence corresponding to the line segment, so that an adjusted pose sequence, namely a second pose sequence, is generated, and then new constraint information, namely preset constraints, can be manually added.
Step 14: and processing the second posture sequence to obtain a third posture sequence so as to ensure that the third posture sequence is continuous.
After the poses in the first pose sequence are adjusted, a continuous pose sequence cannot be formed between the pose sequences corresponding to the selected line segments, and in order to solve the discontinuity caused by pose adjustment, the second pose sequence can be processed to eliminate the discontinuity in the second pose sequence.
Step 15: and performing combined optimization based on the third posture sequence and the received preset constraint to obtain new map data.
After the discontinuity in the second pose sequence is eliminated, joint optimization can be performed according to the constraint information added manually, and the pose sequence and the point cloud data after optimization are generated, namely new map data are generated.
The embodiment provides an SLAM back-end processing method, which comprises the steps of firstly acquiring map data to be processed comprising point cloud data and a pose sequence, and then generating corresponding response data according to manual input; then adjusting the positions of the line segments in the response data so that the adjusted line segments are closest to the point cloud data nearby the line segments; then adjusting the pose sequences corresponding to the adjusted line segments, and processing the adjusted pose sequences to ensure that the processed pose sequences are continuous; then, performing joint optimization based on the processed pose sequence and the manually added preset constraints to generate new map data; the existing processing scheme is prone to online processing, while the map is processed in an offline mode in the embodiment, the map is not related to a mapping method, the pose map can be adjusted through manually selected line segments and added constraints in the later period, then, the back end optimization is carried out, the wrong part in the map is repaired, and the consistency of laser SLAM mapping is improved.
Referring to fig. 2, fig. 2 is a schematic flowchart illustrating another embodiment of a method for processing a SLAM backend provided by the present application, where the method includes:
step 21: receiving map data to be processed, and an operation instruction and an adjustment mode input by a user; response data is generated based on the operation instruction.
The map data to be processed generated in the map building process can be imported, the map data to be processed comprises point cloud data and a pose corresponding to each frame of point cloud data, then a user manually selects each section of point cloud data to be adjusted from the point cloud data, a plurality of line segments are drawn nearby the point cloud data, and a corresponding adjusting mode is set, wherein the adjusting mode comprises overlapping, collinear, parallel or vertical.
Step 22: and adjusting the positions of the line segments by using an expectation maximization method so that the distance between the second point cloud data positioned near the adjusted line segments and the adjusted line segments is closest.
Adjusting the positions of a plurality of line segments drawn by a user by using an Expectation Maximization (EM) method so that point cloud data near the line segments are best fit with the drawn line segments; specifically, the degree of fitting may be determined by the euclidean distance, and a line segment with the smallest sum of the euclidean distances may be selected as the adjusted line segment.
Step 23: and selecting second point cloud data with preset distances of the plurality of line segments after the distance adjustment, acquiring a pose sequence corresponding to the second point cloud data, and recording the pose sequence as a first pose sequence.
And selecting the point cloud data of a plurality of line segments in a certain range after distance adjustment, and acquiring pose sequences corresponding to the point cloud data, wherein the pose sequences are objects needing manual constraint addition.
Step 24: obtaining a pose transformation relation between pose sequences corresponding to the adjusted line segments based on the first pose sequence and the adjustment mode; and adjusting the pose based on the pose transformation relation.
According to the pose corresponding to each segment of point cloud data and the adjustment mode set by the user, the relative pose transformation relation between each segment of pose sequence can be obtained; specifically, a parameter matrix of the t-1 st pose in the first pose sequence can be obtained; multiplying the parameter matrix, the transformation matrix and the change matrix of the t-1 th pose to obtain a parameter matrix of the t-th pose, thereby realizing pose adjustment; the transformation matrix is the transformation relation between the tth pose and the t-1 th pose before pose adjustment, and the transformation matrix is generated by pose adjustment.
Further, before the combined optimization, the pose sequence can be correspondingly processed; taking two line segments as an example, after pose transformation, two selected pose sequences (denoted as X)aAnd Xb) A continuous pose sequence can not be formed between pose sequences, namely at least one place generates pose mutation, discontinuity is generated, and the pose sequence X in pose transformation is assumedaIs fixed, pose sequence XbEarliest pose x intAnd its previous pose xt-1And when sudden change exists, as shown in FIG. 3, the pose after adjustment satisfies the following relation:
xt＝ACxt-1
wherein x istIs the parameter matrix, x, of the t-th poset-1Is the parameter matrix of the t-1 st pose, A is the pose xtAnd pose xt-1A transformation relation before pose adjustment, wherein C is a change matrix generated by pose adjustment; to address the discontinuity introduced by pose adjustment, the following scheme may be employed to process the second sequence of pose positions.
Step 25: and processing the second position sequence by adopting a closed online position and posture chain SLAM method to obtain a third position and posture sequence so as to ensure that the third position and posture sequence is continuous.
Processing poses by using a closed online pose chain SLAM method to eliminate discontinuity in a pose sequence, wherein the core of the closed online pose chain SLAM method is that the pose is processed according to a pose xtThe covariance of each pose in the previous pose sequence is used as a weight to distribute a change matrix C; specifically, a change matrix may be allocated to a preset number of positions before the t-th position, so that the product of the change matrices corresponding to the preset number of positions is equal to the change matrix corresponding to the t-th position, that is:
wherein, UiIs a change matrix corresponding to the ith position before the t-th position, and n is a preset number.
Then processing the preset number of poses before the t-th pose based on the change matrix corresponding to the preset number of poses to obtain a processed pose sequence; the pose x can be made by such processingt-1Each previous pose generates small change, so that the pose xtAnd pose xt-1The pose transformation relation between the pose sequences is restored to an initial state, and then discontinuity in the pose sequences is eliminated; for example, as shown in fig. 4, the dashed circle is a pose sequence in the imported map data to be processed, and the pose sequence X is pose transformedbHas changed position and pose xtAnd pose xt-1The closed online pose chain SLAM passes the pose xtThe previous pose is adjusted to the position of the solid line circle, so that the pose mutation is eliminated.
In a specific embodiment, if data obtained by using a graph optimization mapping method and a final pose graph in a mapping process are imported, closed-loop constraints in an in-situ pose graph can be correspondingly adjusted while the data are processed by a closed-type online pose chain SLAM, otherwise, an error result may occur; because wrong closed-loop constraints may exist in the original pose graph, and the wrong closed-loop constraints are necessarily contradictory to manual constraints, the wrong closed-loop constraints need to be removed before joint optimization.
Step 26: and optimizing the preset constraint by using an optimization method so as to enable a residual error corresponding to the optimized preset constraint to be smaller than a preset residual error, and obtaining adjusted data.
The preset constraint comprises a pose sequence to be adjusted, point cloud data to be adjusted, information of a plurality of adjusted line segments and an adjustment mode; after the manual constraint and the continuous pose sequence are obtained, in order to avoid that the optimization speed is reduced due to excessive constraint and even the global optimal solution cannot be converged, the constraint to be added can be processed and then added into the pose graph so as to carry out combined optimization.
In a specific embodiment, taking the user inputting two line segments as an example, for a given preset constraint h, the preset constraint h is defined as<Pa,Pb,Sa,Sb,Xa,Xb,m>，PaAnd PbFor the end point of the adjusted line section, SaIs PaCorresponding point cloud data, SbIs PbCorresponding point cloud data, XaAnd XbAre respectively SaAnd SbThe corresponding pose, m, is the adjustment mode, and the residual error can be calculated by adopting the following formula:
wherein the content of the first and second substances,
is SaThe ith point cloud data and PaThe euclidean distance between the upper closest points,
is SbThe ith point cloud data and PbEuclidean distance between the upper nearest points, | SaL is SaNumber of midpoint cloud data, | SbL is SbNumber of midpoint cloud data, RaIs PaCorresponding residual error, RbIs PbThe corresponding residual error.
Residual error RaAnd residual error RbWill enhance the pose sequence XaAnd pose sequence XbConnections between internal nodes, residual R in optimizationaAnd residual error RbWill drive the point cloud data SaAnd point cloud data SbEach towards the line segment PaAnd line segment PbMoving, constraining the line segment P by manually set constraintsaAnd line segment PbGeometric relationship between them, and further constrain the point cloud data SaAnd point cloud data SbThe corresponding pose of each section of point cloud data is adjustedThe constraints are added for adjustment for a whole body, and each pose is not independently adjusted.
Step 27: optimizing the adjusted data by using a nonlinear least square method to obtain an optimized pose sequence and optimized point cloud data; and regenerating map data based on the optimized pose sequence and the optimized point cloud data to obtain new map data.
Optimizing the adjusted pose graph by using a nonlinear least square method, and regenerating a new map according to the optimized pose and point cloud data; it will be appreciated that for situations where there are multiple errors in the map, the above steps may be performed multiple times, thereby eliminating errors in the map.
In the embodiment, a pose graph is established in a pose chain mode, generally, the pose graphs generated by back-end optimization are meshed, and in order to eliminate the negative influence of error closed loop on pose optimization, a chain pose graph can be generated according to imported point cloud data and poses corresponding to each frame of point cloud data, namely, a current point cloud frame is only connected with two frames of point cloud data adjacent in time sequence; then manually adding a new closed loop into the pose graph for joint optimization to correct an error part in the map; the off-line map processing is carried out through later-stage manual intervention, the success rate of map building and the consistency of the map are improved, a specific map building method is not depended on, and the data generated by a filtering method and a map optimization method can be subjected to later-stage processing; when the input map data to be processed is data generated by using a map optimization method, errors with smaller errors in a missed map due to errors in subjective judgment can be avoided; in addition, the pose graph in the form of the pose chain can eliminate the influence of error loop on the optimization process, and meanwhile, the closed online pose chain SLAM is used for processing the pose before optimization, so that discontinuous results in a pose sequence can be avoided, and the accuracy of graph construction can be improved.
Referring to fig. 5, fig. 5 is a schematic structural diagram of an embodiment of a SLAM backend processing apparatus provided in the present application, the SLAM backend processing apparatus 50 includes a memory 51 and a processor 52 connected to each other, the memory 51 is used for storing a computer program, and the computer program is used for implementing the SLAM backend processing method in the foregoing embodiment when being executed by the processor 52.
The SLAM back-end processing apparatus 50 provided in this embodiment can add correct closed-loop information on the point cloud data that has completed the map building process by introducing manual adjustment, so as to improve the consistency of the map; in the map processing process, pose processing is performed in advance by combining with a closed online pose chain SLAM method, so that discontinuity of pose sequences before joint optimization is ensured, meanwhile, manual constraints are added to each selected segment of pose sequences as a whole, the situations that optimization speed is reduced and even the optimal solution cannot be converged due to excessive constraints are avoided, and the success rate of map construction can be improved.
Referring to fig. 6, fig. 6 is a schematic structural diagram of an embodiment of a computer-readable storage medium 60 provided in the present application, where the computer-readable storage medium 61 is used for storing a computer program 61, and when the computer program 61 is executed by a processor, the computer program is used for implementing the SLAM backend processing method in the foregoing embodiments.
The computer-readable storage medium 60 may be a server, a usb disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and various media capable of storing program codes.
In the several embodiments provided in the present application, it should be understood that the disclosed method and apparatus may be implemented in other manners. For example, the above-described apparatus embodiments are merely illustrative, and for example, a division of modules or units is merely a logical division, and an actual implementation may have another division, for example, a plurality of units or components may be combined or integrated into another system, or some features may be omitted, or not executed.
Units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the embodiment.
In addition, functional units in the embodiments of the present application may be integrated into one processing unit, or each unit may exist alone physically, or two or more units may be integrated into one unit. The integrated unit can be realized in a form of hardware, and can also be realized in a form of a software functional unit.
The above description is only an example of the present application and is not intended to limit the scope of the present application, and all modifications of equivalent structures and equivalent processes, which are made by the contents of the specification and the drawings, or which are directly or indirectly applied to other related technical fields, are intended to be included within the scope of the present application.
Claims (11)
1. A method of SLAM backend processing, comprising:
receiving map data to be processed and response data, wherein the map data to be processed comprises point cloud data and a pose sequence corresponding to the point cloud data, and the response data comprises information of a plurality of line segments;
adjusting the positions of the line segments based on the information of the line segments and first point cloud data located near the line segments to obtain second point cloud data located near the adjusted line segments and corresponding first position and posture sequences;
adjusting the first position sequence to obtain a second position sequence;
processing the second posture sequence to obtain a third posture sequence so as to enable the third posture sequence to be continuous;
and performing joint optimization based on the third posture sequence and the received preset constraint to obtain new map data.
2. The SLAM backend processing method according to claim 1, wherein the step of adjusting the positions of the plurality of line segments based on the information of the plurality of line segments and first point cloud data located in the vicinity of the plurality of line segments comprises:
and adjusting the positions of the line segments by using an expectation maximization method so that the distance between the second point cloud data positioned near the adjusted line segments and the adjusted line segments is closest.
3. The SLAM backend processing method of claim 2, wherein the step of adjusting the first sequence of bit positions to obtain a second sequence of bit positions comprises:
selecting second point cloud data with preset distances from the adjusted line segments, acquiring a pose sequence corresponding to the second point cloud data, and recording the pose sequence as the first pose sequence;
obtaining a pose transformation relation between pose sequences corresponding to the adjusted line segments based on the first pose sequence and an adjustment mode;
and adjusting the pose based on the pose transformation relation.
4. The SLAM backend processing method according to claim 3, wherein the step of performing pose adjustment based on the pose transformation relationship includes:
acquiring a parameter matrix of the t-1 st pose in the first pose sequence;
multiplying the parameter matrix, the transformation matrix and the change matrix of the t-1 th pose to obtain a parameter matrix of the t-th pose;
wherein the transformation matrix is a transformation relation between the tth pose and the t-1 th pose before pose adjustment, and the transformation matrix is generated by pose adjustment.
5. The SLAM back-end processing method of claim 4, wherein the step of processing the second sequence of pose positions to obtain a third sequence of pose positions such that the third sequence of pose positions is continuous comprises:
and processing the second position and posture sequence by adopting a closed online position and posture chain SLAM method.
6. The SLAM back-end processing method of claim 5, wherein the step of processing the second sequence of position poses with a closed online pose chain SLAM method comprises:
distributing a change matrix for a preset number of position postures before the t-th position posture so that the product of the change matrices corresponding to the preset number of position postures is equal to the change matrix corresponding to the t-th position posture;
and processing the preset number of posture positions before the t-th posture position based on the change matrix corresponding to the preset number of posture positions to obtain the third posture position sequence.
7. The SLAM backend processing method according to claim 3, wherein the step of performing joint optimization based on the third pose sequence and the received preset constraints to obtain new map data comprises:
optimizing the preset constraint by using an optimization method so that a residual error corresponding to the optimized preset constraint is smaller than a preset residual error to obtain adjusted data, wherein the preset constraint comprises the pose sequence to be adjusted, the point cloud data to be adjusted, the information of the adjusted line segments and the adjustment mode;
optimizing the adjusted data by utilizing a nonlinear least square method to obtain an optimized pose sequence and optimized point cloud data;
and regenerating map data based on the optimized pose sequence and the optimized point cloud data to obtain the new map data.
8. The SLAM back-end processing method of claim 7,
the residual error is calculated using the following formula:
wherein the predetermined constraint is denoted as h ═ h<Pa,Pb,Sa,Sb,Xa,Xb,m>，PaAnd PbIs the end point of the adjusted line segment, SaIs PaCorresponding point cloud data, SbIs PbCorresponding point cloud data, XaAnd XbAre respectively SaAnd SbThe corresponding pose, m, is the adjustment mode,
is SaThe ith point cloud data and PaThe euclidean distance between the upper closest points,
is SbThe ith point cloud data and PbEuclidean distance between the upper nearest points, | SaL is SaNumber of midpoint cloud data, | SbL is SbNumber of midpoint cloud data, RaIs PaCorresponding residual error, RbIs PbThe corresponding residual error.
9. The SLAM backend processing method according to claim 1, wherein the step of comparing the information based on the plurality of line segments with first point cloud data located in the vicinity of the plurality of line segments is preceded by:
receiving an operation instruction input by a user, and generating the response data based on the operation instruction;
receiving an adjustment mode input by a user, wherein the adjustment mode comprises overlapping, collinear, parallel or perpendicular.
10. A SLAM back-end processing apparatus comprising a memory and a processor connected to each other, wherein the memory is configured to store a computer program, and the computer program is configured to implement the SLAM back-end processing method according to any one of claims 1 to 9 when executed by the processor.
11. A computer-readable storage medium storing a computer program for implementing the SLAM back-end processing method of any one of claims 1-9 when executed by a processor.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010778658.0A CN112052300A (en) | 2020-08-05 | 2020-08-05 | SLAM back-end processing method, device and computer readable storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010778658.0A CN112052300A (en) | 2020-08-05 | 2020-08-05 | SLAM back-end processing method, device and computer readable storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112052300A true CN112052300A (en) | 2020-12-08 |
Family
ID=73601394
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010778658.0A Pending CN112052300A (en) | 2020-08-05 | 2020-08-05 | SLAM back-end processing method, device and computer readable storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112052300A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113674411A (en) * | 2021-07-29 | 2021-11-19 | 浙江大华技术股份有限公司 | Pose graph adjustment-based graph establishing method and related equipment |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107272673A (en) * | 2017-05-18 | 2017-10-20 | 中山大学 | SLAM rear ends track optimizing method based on pose chain model |
| WO2019169540A1 (en) * | 2018-03-06 | 2019-09-12 | 斯坦德机器人（深圳）有限公司 | Method for tightly-coupling visual slam, terminal and computer readable storage medium |
| CN110675483A (en) * | 2019-07-17 | 2020-01-10 | 电子科技大学 | Dense vision SLAM-based rapid reconstruction method for three-dimensional map of unmanned aerial vehicle |
| CN110866927A (en) * | 2019-11-21 | 2020-03-06 | 哈尔滨工业大学 | Robot positioning and composition method based on EKF-SLAM algorithm combined with dotted line characteristics of foot |
| US20200089236A1 (en) * | 2016-12-26 | 2020-03-19 | Bayerische Motoren Werke Aktiengesellschaft | Method and Apparatus for Optimizing Scan Data and Method and Apparatus for Correcting Trajectory |
| CN111325796A (en) * | 2020-02-28 | 2020-06-23 | 北京百度网讯科技有限公司 | Method and apparatus for determining pose of vision device |
-
2020
- 2020-08-05 CN CN202010778658.0A patent/CN112052300A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20200089236A1 (en) * | 2016-12-26 | 2020-03-19 | Bayerische Motoren Werke Aktiengesellschaft | Method and Apparatus for Optimizing Scan Data and Method and Apparatus for Correcting Trajectory |
| CN107272673A (en) * | 2017-05-18 | 2017-10-20 | 中山大学 | SLAM rear ends track optimizing method based on pose chain model |
| WO2019169540A1 (en) * | 2018-03-06 | 2019-09-12 | 斯坦德机器人（深圳）有限公司 | Method for tightly-coupling visual slam, terminal and computer readable storage medium |
| CN110675483A (en) * | 2019-07-17 | 2020-01-10 | 电子科技大学 | Dense vision SLAM-based rapid reconstruction method for three-dimensional map of unmanned aerial vehicle |
| CN110866927A (en) * | 2019-11-21 | 2020-03-06 | 哈尔滨工业大学 | Robot positioning and composition method based on EKF-SLAM algorithm combined with dotted line characteristics of foot |
| CN111325796A (en) * | 2020-02-28 | 2020-06-23 | 北京百度网讯科技有限公司 | Method and apparatus for determining pose of vision device |
Non-Patent Citations (6)
| Title |
|---|
| ZIQUAN LAN: "Robust Point Cloud Based Reconstruction of Large-Scale Outdoor Scenes", 2019 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR), 9 January 2020 (2020-01-09) * |
| 张毅;沙建松;: "基于图优化的移动机器人视觉SLAM", 智能系统学报, no. 02, 2 July 2017 (2017-07-02) * |
| 田伟: "仓储环境中移动机器人拓扑地图构建", 中国硕士学位论文全文数据库 信息科技辑, 15 January 2020 (2020-01-15) * |
| 禹鑫D;朱熠琛;詹益安;欧林林;: "SLAM过程中的机器人位姿估计优化算法研究", 高技术通讯, no. 08, 15 August 2018 (2018-08-15) * |
| 翁秀玲;王云峰;吴炜;郭东辉;: "基于局内点分布的图优化方法", 厦门大学学报(自然科学版), no. 02, 30 September 2017 (2017-09-30) * |
| 蒋林;夏旭洪;韩璐;邱存勇;张泰;宋杰;: "一种点线特征融合的双目同时定位与地图构建方法", 科学技术与工程, no. 12, 28 April 2020 (2020-04-28) * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113674411A (en) * | 2021-07-29 | 2021-11-19 | 浙江大华技术股份有限公司 | Pose graph adjustment-based graph establishing method and related equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN101996420B (en) | Information processing device, information processing method and program | |
| US11263539B2 (en) | Distributed machine learning method and system | |
| US11587356B2 (en) | Method and device for age estimation | |
| US20100053451A1 (en) | Apparatus and method for frame interpolation based on accurate motion estimation | |
| CN112947084B (en) | Model unknown multi-agent consistency control method based on reinforcement learning | |
| CN113792730A (en) | Method and device for correcting document image, electronic equipment and storage medium | |
| CN111079695A (en) | Human body key point detection and self-learning method and device | |
| CN111210475A (en) | Map updating method and device | |
| CN111144494A (en) | Object detection model training method, object detection device, object detection equipment and object detection medium | |
| CN110378942A (en) | Barrier identification method, system, equipment and storage medium based on binocular camera | |
| CN112232426A (en) | Training method, device and equipment of target detection model and readable storage medium | |
| CN112052300A (en) | SLAM back-end processing method, device and computer readable storage medium | |
| CN111178261A (en) | Face detection acceleration method based on video coding technology | |
| US20240005595A1 (en) | Three-dimensional reconstruction method, system, and storage medium | |
| CN109284062A (en) | A kind of touch data processing method, device, terminal and medium | |
| WO2024077767A1 (en) | Learning model-oriented coding decision processing method and apparatus, and device | |
| US11688175B2 (en) | Methods and systems for the automated quality assurance of annotated images | |
| CN109308360A (en) | Channel parameter finding method, device, storage medium and computer equipment | |
| CN113887290A (en) | Monocular 3D detection method and device, electronic equipment and storage medium | |
| JP2003178310A (en) | Object image tracking method and program | |
| CN114387197A (en) | Binocular image processing method, device, equipment and storage medium | |
| CN104243844A (en) | Image processor, image processing method and electronic equipment | |
| CN115408549B (en) | Workpiece point cloud filtering method and device, computer readable medium and electronic equipment | |
| CN114025079B (en) | Image quality optimization parameter processing method, device and system | |
| CN117600690B (en) | Laser galvanometer scanning system and calibration method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |