CN108981728B - Intelligent vehicle navigation map building method - Google Patents
Intelligent vehicle navigation map building method Download PDFInfo
- Publication number
- CN108981728B CN108981728B CN201810834740.3A CN201810834740A CN108981728B CN 108981728 B CN108981728 B CN 108981728B CN 201810834740 A CN201810834740 A CN 201810834740A CN 108981728 B CN108981728 B CN 108981728B
- Authority
- CN
- China
- Prior art keywords
- terrain
- data
- vehicle
- information
- road
- 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.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
- G01C21/00—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00
- G01C21/26—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 specially adapted for navigation in a road network
- G01C21/28—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 specially adapted for navigation in a road network with correlation of data from several navigational instruments
- G01C21/30—Map- or contour-matching
- G01C21/32—Structuring or formatting of map data
Abstract
The invention relates to an intelligent vehicle navigation map building method, which comprises the following steps: the method comprises the following steps that a collection vehicle collects original data under typical terrain working conditions, and a terrain identification SVM model is trained; identifying the terrain working condition of the collected vehicle on a running road by adopting a trained terrain identification SVM model to obtain the terrain working condition information of the road network; and matching the road position information and the topographic condition information to establish the electronic map with the topographic condition information. The intelligent vehicle navigation map established by the invention can provide the intelligent vehicle with the road surface bump condition, and can improve the safety of tracking and driving of the unmanned vehicle; more choices are provided for human drivers and auxiliary driving systems, and the driving stability and safety are ensured; the driver is helped to select a driving mode and a driving route which are more suitable for driving, and the driving safety and the driving comfort are improved; the navigation map may also provide reliable data entry for energy recovery management techniques, facilitating efficient application of the techniques.
Description
Technical Field
The invention relates to the technical field of electronic maps, in particular to an intelligent vehicle navigation map establishing method.
Background
The electronic map which is widely used at present is a product generated along with the rapid development of computer science technology in recent decades, and the common electronic map which is applied at present can only provide basic information such as positions, pictures and the like; this type of electronic map does not contain information about the terrain conditions, and has not been able to meet the demand of vehicles, especially intelligent navigation vehicles, for electronic maps.
Disclosure of Invention
In view of the above analysis, the present invention aims to provide an intelligent vehicle navigation map building method, which adds topographic information to an electronic map to solve the shortcomings and drawbacks of the existing map information.
The purpose of the invention is mainly realized by the following technical scheme:
an intelligent vehicle navigation map building method comprises the following steps:
acquiring original data under typical terrain working conditions, and training a terrain identification SVM model;
identifying the terrain working condition of the collected vehicle driving road by adopting a trained terrain identification SVM model to obtain the terrain working condition information of the road;
and matching the position information of the road with the topographic condition information to establish an intelligent vehicle navigation map.
Further, the collecting raw data under typical terrain conditions and training a terrain recognition SVM model includes:
1) collecting the road running of a vehicle under various typical terrain working conditions, and collecting original data including vehicle attitude information;
2) according to different topographic conditions, data extraction and labeling processing are carried out on the original data to obtain labeled extracted data;
3) performing unitization processing on the extracted data to obtain segmented data;
4) carrying out normalization processing on the segmented data;
5) extracting and screening attribute features for SVM modeling, and dividing the data subjected to normalization processing into a training sample set and a test sample set;
6) training the SVM model by using the training sample set data to obtain a terrain classification SVM model; and testing the classification effect of the terrain classification SVM model by using the test sample set data, and optimizing the model parameters.
Further, the data extraction and tagging of the raw data includes:
1) according to the actual driving condition of the collected vehicle, roughly determining the time periods of the original data of different terrain working conditions;
2) determining starting time stamps and ending time stamps of different terrain working condition data;
3) and intercepting continuous original data according to the time stamp, and marking the intercepted data according to the actual topographic working condition to obtain tagged extracted data.
Further, the criteria of the unitization process include:
taking the intersection point of the time variation curve of the pitch angle and a straight line with the pitch angle equal to 0 as a segmentation point of data;
if the pitch angle change trends of the left side and the right side of a certain data point are different, and the maximum value of the absolute value of the change rate of the two sides is larger than a critical value, the point is determined as a segmented point;
and checking the segmented points which are segmented, and if the change rate of the pitch angle at the segmented points along with the time exceeds a critical value, canceling the segmented points.
Further, the normalization processing adopts a linear function normalization method.
Further, the method for extracting and screening the attribute features for SVM modeling comprises the following steps:
1) calculating the weight of all attribute features of the collected original data by adopting a Relief-F algorithm, sequencing all attribute features according to the weight, and enabling the weight to be larger than a critical value WαForm a new set D of attribute featuresRF；
2) Using CON selection method and according to consistency of attribute subset and category, for the set DRFSearch for feature subset DCON；
3) Adopting CFS selection method, and according to correlation characteristics between different attribute characteristics, performing D on the setRFSearching for corresponding feature subsets DCFS；
4) Merging feature subsets DCON、DCFSAnd obtaining an attribute feature set for SVM modeling.
Furthermore, in the model training process, the training sample set data is used for carrying out modeling training of terrain identification on the kernel functions of the SVM model respectively, and the kernel functions with high observation identification accuracy and few iteration times are selected as the kernel functions of the terrain identification model.
Further, the parameter optimization method adopts a particle swarm algorithm and a cross validation method.
Further, the step of matching the position information and the terrain condition information of the road to establish an intelligent vehicle navigation map comprises the following steps:
s21, collecting the running of the vehicle on the road, and collecting the position information and the posture information of the vehicle;
s22, performing terrain identification on the attitude information by using the trained terrain identification SVM model to obtain terrain working condition information;
step S23, carrying out data matching on the terrain condition information and the vehicle position information;
and step S24, making the road into the electronic map according to the determined road information including the type, length and start-stop position of the terrain condition.
Further, the data matching is achieved by corresponding the vehicle position information and the time stamp of the terrain condition information. The invention has the following beneficial effects:
the intelligent vehicle navigation map established by the invention can provide the road surface bump condition for the intelligent vehicle, is an important technical way for improving the safety of unmanned vehicle track planning, and greatly improves the safety of unmanned vehicle tracking driving;
for the advanced auxiliary driving field, the terrain condition information in the navigation map also provides more choices for human drivers and auxiliary driving systems, and the driving stability and safety are ensured;
for a vehicle driven by a common driver, the terrain condition information in the navigation map can help the driver to select a driving mode and a driving route which are more suitable for driving, so that the driving safety and the driving comfort are improved;
the navigation map may also provide reliable data entry for energy recovery management techniques, facilitating efficient application of the techniques.
Additional features and advantages of the invention will be set forth in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. The objectives and other advantages of the invention will be realized and attained by the structure particularly pointed out in the written description and claims hereof as well as the appended drawings.
Drawings
The drawings are only for purposes of illustrating particular embodiments and are not to be construed as limiting the invention, wherein like reference numerals are used to designate like parts throughout.
FIG. 1 is a flowchart of a method for creating an intelligent vehicle navigation map according to an embodiment of the present invention.
Detailed Description
The preferred embodiments of the present invention will now be described in detail with reference to the accompanying drawings, which form a part hereof, and which together with the embodiments of the invention serve to explain the principles of the invention.
The embodiment of the invention provides an intelligent vehicle navigation map building method, which builds an intelligent vehicle navigation map by adding topographic information to an electronic map.
As shown in fig. 1, the method comprises the following steps:
s1, collecting original data of the vehicle under typical terrain working conditions, and training a terrain recognition SVM model;
the acquisition vehicle comprises an inertial navigation system, a GPS positioning device, a camera and other vehicle-mounted sensing sensors and is used for acquiring real-time attitude information, position information and environmental picture information of the vehicle;
the real-time attitude information of the vehicle is output through a vehicle-mounted inertial navigation system and comprises vehicle running information such as a course angle, a pitch angle, a roll angle, a vehicle speed and the like of the vehicle;
the real-time position information of the vehicle is output by a GPS positioning device;
the environment picture information is shot by the vehicle-mounted camera, and in the acquisition process, the vehicle-mounted camera continuously shoots the vehicle running environment to serve as comparison data for judging the terrain working condition afterwards.
Specifically, the training of the terrain recognition SVM model comprises the following steps:
s11, collecting the running of the vehicle on a road containing various typical terrain working conditions, and collecting original data including vehicle attitude information;
when the driving route of the collection vehicle is selected, a route containing enough typical terrain working condition types is selected to meet the requirement of model training.
Step S12, according to different terrain conditions, data extraction and tagging are carried out on the original data to obtain tagged extracted data;
an SVM model used in the process of performing terrain recognition is a typical supervised learning algorithm, and should be trained using labeled data. Labeling is to label the real working condition of the data, and record the data in the form of a label as a part of input when training the support vector machine model. In the data collected in the actual road section, the data are often collected continuously, so that a tester needs to extract and divide the data according to different terrain working conditions after testing.
The specific data extraction and labeling processing method comprises the following steps:
firstly, according to the actual running condition of a collection vehicle, roughly determining the time periods of original data of different terrain working conditions;
optionally, the approximate time of the terrain condition data segment to be extracted can be determined according to an acquisition record table recording the time nodes of the acquisition process;
determining starting time stamps and ending time stamps of different terrain working condition data;
in the data acquisition process of the acquisition vehicle, the vehicle-mounted camera continuously shoots the running environment pictures, numbers the pictures and determines the time stamp.
After determining an approximate time period for extracting data, extracting a picture of the vehicle-mounted camera in the vicinity of the extraction time period;
determining the initial photo serial number and the tail photo serial number of the extracted data corresponding to the terrain working condition according to the environmental information reflected by the photos;
according to the serial numbers of the initial and the last photos, the initial and the last timestamps of the extracted target data segment are found and recorded in combination with the timestamps;
and thirdly, intercepting continuous original data according to the time stamps, and marking the intercepted data according to actual terrain working conditions to obtain tagged extracted data.
Step S13, unitizing the extracted data to obtain segmented data;
finding the optimal data unitization strategy is the necessary route for terrain identification. The data unitization strategy needs to make the data units as small as possible in time span so as to reduce the inherent delay of the terrain identification algorithm, and also needs to ensure that the information contained in each data unit can represent the corresponding actual terrain working condition so as to ensure the terrain identification effect. In combination with this principle, the data is unitized according to the following criteria:
and segmenting the data at the intersection point every time when a time variation curve of the pitch angle in the extracted data intersects with a straight line with the pitch angle equal to 0.
Checking the segmented points, and if the change rate of the pitch angle from the segmented points along with the time exceeds a critical value (the critical value is set to be 0.05), canceling the segmented points.
Thirdly, if the pitch angle change trends of the left side and the right side of a certain data point are different (namely, the pitch angle changes gradually increase and gradually decrease), and the maximum value of the absolute value of the change rate of the two sides is larger than a critical value, the point is determined as a segmentation point (the standard aims to take the peak and the valley point of the change as the segmentation point).
And performing segmentation operation on the data by using the standard, recording time stamps at all segmentation points as a basis for segmenting the data, and performing unitization processing on the data segments obtained after data extraction to obtain a series of data units.
Step S14, normalization processing is carried out on the segmented data;
the data normalization method is to perform dimensionless processing on the data so that the waveform characteristics of the data become smooth. And the data can be restricted in a specific range, and the data with the data value higher than the normalization upper limit or lower than the normalization lower limit can be adjusted to be kept in the specific range. In addition, the data preprocessing process also enables the data to accord with the kernel radius condition of the Gaussian kernel function, so that the optimal strategy function can be optimized correspondingly. A linear function normalization method is used here.
Also called the most value normalization method, uses linear function to map the original data into the range, and adopts formula
And (4) converting, wherein max and min are respectively the maximum value and the minimum value of the data. Since the method is mapped by using a linear function, the method can realize equal scaling of data.
Through normalization, the identification accuracy can be improved, and the time consumption in the training process can be reduced.
S15, extracting and screening attribute characteristic parameters for SVM modeling, and dividing the normalized data into a training sample set and a test sample set;
in the process of terrain identification modeling, if all attribute features are used, the computation amount in the model training process and the model classification process is greatly increased, the training efficiency and the classification accuracy of the model are seriously influenced, and the model training efficiency is negatively influenced due to the mutual coupling relationship among different attribute features; therefore, it is desirable to construct a good attribute feature set that contains abundant and sufficient class information to allow different classes of data to be distinguished and at the same time minimize the coupling between attribute features.
Particularly, the attribute feature extraction and screening method for SVM modeling adopted in the embodiment of the present invention includes:
calculating weights of all attribute characteristics in acquired attitude information by adopting a Relief-F algorithm, sequencing all attribute characteristics according to the weights, and enabling the weights to be larger than a critical value WαForm a new set D of attribute featuresRF(ii) a Critical value WαAnd setting according to requirements.
Secondly, a CON (consistent-based method) selection method based on the characteristic consistency measurement is adopted, and the set D is subjected to consistency of the attribute subset and the categoryRFSearch for feature subset DCON。
Adopting a CFS (correlation-based feature selection) method based on feature correlation measurement, and according to correlation characteristics among different attribute features, performing selection on the set DRFSearching for corresponding feature subsets DCFS。
Merging feature subsets DCON、DCFSAnd obtaining an attribute feature set for SVM modeling.
The feature set is generated by considering the importance of each attribute as a single attribute, and considering the consistency and relevance among the attribute features in the overall view of the set of the attribute features.
S16, training the SVM model by using the training sample set data to obtain a terrain classification SVM model; and testing the classification effect of the terrain classification SVM model by using the test sample set data, and optimizing the model parameters.
The core idea of SVM is to find the best classification hyperplane so that positive and negative examples are separated by it and the positive and negative examples are as far away from the hyperplane as possible. The idea of dimension raising is used in the process of finding the optimal classification hyperplane, the process of dimension raising of the sample points is mainly realized by using a kernel function, and any SVM classifier cannot guarantee 100% accuracy, so a punishment parameter is often set to control the classification effect of the classifier.
Particularly, in the model training process, the modeling training of terrain identification is respectively carried out on the current mainstream SVM kernel functions line, polymodal, radial basis function and sigmoid by using training sample set data, and kernel functions with high observation and identification accuracy and less iteration times are selected from the current mainstream SVM kernel functions line, polymodal, radial basis function and sigmoid to serve as the kernel functions of the terrain identification model; in this embodiment, the most suitable fast terrain classification SVM kernel is the radial basis function.
After the kernel function type is selected, selecting a proper kernel function parameter g and a proper penalty parameter C can also have positive influence on the training efficiency and the classification accuracy;
specifically, the embodiment of the present invention performs parameter optimization by using two methods. The first is to use particle swarm optimization to select kernel function parameters, and the other is to use cross validation method to select punishment parameters; the kernel function parameter g is 0.02, and the penalty parameter C is 2, which is the most suitable relevant parameter of the fast terrain classification SVM model of the invention.
After the processes are gradually carried out and the terrain recognition SVM model is established, the terrain working condition can be recognized.
S2, identifying the terrain condition of the road where the collection vehicle runs by adopting the trained terrain identification SVM model to obtain the terrain condition information of the road; and matching the position information and the topographic condition information of the road to manufacture the electronic map.
The method specifically comprises the following steps:
s21, collecting the running of the vehicle on the road, and collecting the position information and the posture information of the vehicle;
the method comprises the steps that an acquisition vehicle runs on a road, and in the process of acquiring the road network information of the electronic map, information is stored as position information and a corresponding timestamp, wherein the position information is a position information part of the road network information;
s22, performing terrain identification on the attitude information through the trained terrain identification SVM model to obtain terrain working condition information;
carrying out terrain identification by using a trained terrain identification SVM model according to the acquired attitude information, recording the identification result, and storing the information as a classification result with terrain identification and a corresponding timestamp, which is a terrain information part of the road network information;
step S23, carrying out data matching on the terrain condition information and the vehicle position information;
the two kinds of road network information are sent to a user side for manufacturing an electronic map by an acquisition vehicle, and the user side loads the two kinds of road network information; and corresponding the time stamps of the position information part and the topographic information part of the road network information to realize the correspondence between topographic condition information and position information, so that the environmental topographic information is matched with the vehicle position information.
And step S24, making the road into the electronic map according to the determined road information including the type, length and start-stop position of the terrain condition.
The user side for manufacturing the electronic map also determines the starting and stopping point positions of the terrain working conditions according to the position information corresponding to the starting and stopping points of the terrain working conditions related in the terrain working condition information, calculates the length of the terrain working conditions according to the position information, adds road information including the types, the lengths and the starting and stopping positions of the terrain working conditions into the electronic map, and repeats the steps to continuously manufacture the electronic map until the collection vehicle finishes collecting and stops running.
Sending the road network information to a navigation user needing the function of the electronic map containing the topographic information to perform map navigation; the navigation user uses the electronic map containing the terrain information to navigate, so that the navigation capable of providing the terrain information can be realized without using an inertial navigation system with high cost.
In conclusion, the intelligent vehicle navigation map established by the embodiment of the invention can provide the intelligent vehicle with the road surface bump condition, is an important technical way for improving the safety of unmanned vehicle track planning, and greatly improves the safety of unmanned vehicle tracking driving; for the advanced auxiliary driving field, the terrain condition information in the navigation map also provides more choices for human drivers and auxiliary driving systems, and the driving stability and safety are ensured; for a vehicle driven by a common driver, the terrain condition information in the navigation map can help the driver to select a driving mode and a driving route which are more suitable for driving, so that the driving safety and the driving comfort are improved; the navigation map may also provide reliable data entry for energy recovery management techniques, facilitating efficient application of the techniques.
Those skilled in the art will appreciate that all or part of the flow of the method implementing the above embodiments may be implemented by a computer program, which is stored in a computer readable storage medium, to instruct related hardware. The computer readable storage medium is a magnetic disk, an optical disk, a read-only memory or a random access memory.
The above description is only for the preferred embodiment of the present invention, but the scope of the present invention is not limited thereto, and any changes or substitutions that can be easily conceived by those skilled in the art within the technical scope of the present invention are included in the scope of the present invention.
Claims (8)
1. An intelligent vehicle navigation map building method is characterized by comprising the following steps:
acquiring original data under typical terrain working conditions, and training a terrain identification SVM model;
identifying the terrain working condition of the collected vehicle driving road by adopting a trained terrain identification SVM model to obtain the terrain working condition information of the road;
matching the position information of the road with the topographic condition information to establish an intelligent vehicle navigation map;
the collecting of the original data under the typical terrain working condition and the training of the terrain recognition SVM model comprises the following steps:
1) collecting the road running of a vehicle under various typical terrain working conditions, and collecting original data including vehicle attitude information;
2) according to different topographic conditions, data extraction and labeling processing are carried out on the original data to obtain labeled extracted data;
3) performing unitization processing on the extracted data to obtain segmented data;
4) carrying out normalization processing on the segmented data;
5) extracting and screening attribute features for SVM modeling, and dividing the data subjected to normalization processing into a training sample set and a test sample set;
6) training the SVM model by using the training sample set data to obtain a terrain classification SVM model; testing the classification effect of the terrain classification SVM model by using test sample set data, and optimizing model parameters;
the data extraction and labeling processing of the original data comprises the following steps:
1) according to the actual driving condition of the collected vehicle, roughly determining the time periods of the original data of different terrain working conditions;
2) determining starting time stamps and ending time stamps of different terrain working condition data;
3) and intercepting continuous original data according to the time stamp, and marking the intercepted data according to the actual topographic working condition to obtain tagged extracted data.
2. The map building method according to claim 1, wherein the criteria for the unitization process include:
taking the intersection point of the time variation curve of the pitch angle and a straight line with the pitch angle equal to 0 as a segmentation point of data;
if the pitch angle change trends of the left side and the right side of a certain data point are different, and the maximum value of the absolute value of the change rate of the two sides is larger than a critical value, the point is determined as a segmented point;
and checking the segmented points which are segmented, and if the change rate of the pitch angle at the segmented points along with the time exceeds a critical value, canceling the segmented points.
3. The map building method according to claim 1, wherein the normalization process employs a linear function normalization method.
4. The map building method according to claim 1,
the method for extracting and screening the attribute features for SVM modeling comprises the following steps:
1) calculating the weight of all attribute features of the collected original data by adopting a Relief-F algorithm, sequencing all attribute features according to the weight, and enabling the weight to be larger than a critical value WαForm a new set D of attribute featuresRF；
2) Using CON selection method and according to consistency of attribute subset and category, for the set DRFSearch for feature subset DCON；
3) Adopting CFS selection method, and according to correlation characteristics between different attribute characteristics, performing D on the setRFSearching for corresponding feature subsets DCFS；
4) Merging feature subsets DCON、DCFSAnd obtaining an attribute feature set for SVM modeling.
5. The map building method according to claim 1, wherein in the model training process, the training sample set data is used for respectively carrying out modeling training of terrain identification on the kernel functions of the SVM model, and the kernel functions with high observation identification accuracy and few iteration times are selected as the kernel functions of the terrain identification model.
6. The map building method according to claim 1, wherein the parameter optimization method employs a particle swarm algorithm and a cross validation method.
7. The map building method according to claim 1, wherein the matching of the position information and the topographic condition information of the road to build the intelligent vehicle navigation map comprises:
s21, collecting the running of the vehicle on the road, and collecting the position information and the posture information of the vehicle;
s22, performing terrain identification on the attitude information by using the trained terrain identification SVM model to obtain terrain working condition information;
step S23, carrying out data matching on the terrain condition information and the vehicle position information;
and step S24, making the road into the electronic map according to the determined road information including the type, length and start-stop position of the terrain condition.
8. The map building method according to claim 7, wherein the data matching is performed by correlating time stamps of the vehicle position information and the terrain condition information.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810834740.3A CN108981728B (en) | 2018-07-25 | 2018-07-25 | Intelligent vehicle navigation map building method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810834740.3A CN108981728B (en) | 2018-07-25 | 2018-07-25 | Intelligent vehicle navigation map building method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108981728A CN108981728A (en) | 2018-12-11 |
| CN108981728B true CN108981728B (en) | 2020-11-10 |
Family
ID=64551452
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810834740.3A Active CN108981728B (en) | 2018-07-25 | 2018-07-25 | Intelligent vehicle navigation map building method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108981728B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110377032B (en) * | 2019-07-08 | 2022-05-27 | 佳木斯市奥义智能科技有限公司 | Semantic map lower robot sloping field motion control method and device and storage medium |
| CN112785909B (en) * | 2019-11-06 | 2022-12-02 | 北京地平线机器人技术研发有限公司 | Determination method and device for occupying grid map |
| CN111103885A (en) * | 2019-12-31 | 2020-05-05 | 深圳前海达闼云端智能科技有限公司 | Robot control method, robot control device, electronic device, and storage medium |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102221363A (en) * | 2011-04-12 | 2011-10-19 | 东南大学 | Fault-tolerant combined method of strapdown inertial integrated navigation system for underwater vehicles |
| CN103645480A (en) * | 2013-12-04 | 2014-03-19 | 北京理工大学 | Geographic and geomorphic characteristic construction method based on laser radar and image data fusion |
| EP3340130A1 (en) * | 2016-12-23 | 2018-06-27 | Hexagon Technology Center GmbH | Method for prediction of soil and/or plant condition |
-
2018
- 2018-07-25 CN CN201810834740.3A patent/CN108981728B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102221363A (en) * | 2011-04-12 | 2011-10-19 | 东南大学 | Fault-tolerant combined method of strapdown inertial integrated navigation system for underwater vehicles |
| CN103645480A (en) * | 2013-12-04 | 2014-03-19 | 北京理工大学 | Geographic and geomorphic characteristic construction method based on laser radar and image data fusion |
| EP3340130A1 (en) * | 2016-12-23 | 2018-06-27 | Hexagon Technology Center GmbH | Method for prediction of soil and/or plant condition |
Non-Patent Citations (4)
| Title |
|---|
| Christian Weiss 等.Vibration-based Terrain Classification Using Support Vector Machines.《Proceedings of the 2006 IEEE/RSJ 》.2006,第4429-4434页. * |
| Vibration-based Terrain Classification Using Support Vector Machines;Christian Weiss 等;《Proceedings of the 2006 IEEE/RSJ 》;20061015;4429-4434页 * |
| 基于振动采用支持向量机方法的移动机器人地形分类;李强等;《机器人》;20121115(第06期);660-667页 * |
| 基于视觉采用词袋模型的移动机器人地形分类算法设计;孙玉超等;《医疗卫生装备》;20170215(第02期);114-121页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108981728A (en) | 2018-12-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Li et al. | Line-cnn: End-to-end traffic line detection with line proposal unit | |
| Chen et al. | Deep integration: A multi-label architecture for road scene recognition | |
| CN108830188B (en) | Vehicle detection method based on deep learning | |
| CN107563372B (en) | License plate positioning method based on deep learning SSD frame | |
| Garcia-Fidalgo et al. | Vision-based topological mapping and localization methods: A survey | |
| Cieslewski et al. | Point cloud descriptors for place recognition using sparse visual information | |
| Xin et al. | Localizing discriminative visual landmarks for place recognition | |
| Mousavian et al. | Semantically guided location recognition for outdoors scenes | |
| CN108981728B (en) | Intelligent vehicle navigation map building method | |
| CN104615986A (en) | Method for utilizing multiple detectors to conduct pedestrian detection on video images of scene change | |
| CN103279738A (en) | Automatic identification method and system for vehicle logo | |
| Singh et al. | Acquiring semantics induced topology in urban environments | |
| CN111767905A (en) | Improved image method based on landmark-convolution characteristics | |
| CN113705570A (en) | Few-sample target detection method based on deep learning | |
| Singh et al. | Semantically guided geo-location and modeling in urban environments | |
| Hasan et al. | Traffic sign recognition system (TSRS): SVM and convolutional neural network | |
| CN104778699A (en) | Adaptive object feature tracking method | |
| Han et al. | A novel loop closure detection method with the combination of points and lines based on information entropy | |
| Gong et al. | A two-level framework for place recognition with 3D LiDAR based on spatial relation graph | |
| CN115311229A (en) | Laser radar-based pavement disease detection and classification method and system and storage medium | |
| Cai et al. | Patch-NetVLAD+: Learned patch descriptor and weighted matching strategy for place recognition | |
| Hou et al. | Use of roadway scene semantic information and geometry-preserving landmark pairs to improve visual place recognition in changing environments | |
| Dai et al. | Road surface detection and recognition for route recommendation | |
| CN113723558A (en) | Remote sensing image small sample ship detection method based on attention mechanism | |
| CN103559492A (en) | Car logo recognition device and 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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |